Report No: AUS0000414 . Afghanistan Education and Skills Programmatic ASA Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan . August 2018 . Education GP . . . © 2017 The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved This work is a product of the staff of The World Bank. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of The World Bank 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. Attribution—Please cite the work as follows: “World Bank. (2018). Teacher Policy Study in Afghanistan. © World Bank.� All 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. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 2 Teacher Policy Study Towards a more Equitable and Efficient Use of Education Resources in Afghanistan The World Bank August 2018 Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 3 Table of Contents ACKNOWLEDGEMENT 6 LIST OF TABLES, FIGURES, MAPS, AND BOXES 7 ACRONYMS & GLOSSARY 9 EXECUTIVE SUMMARY 11 1 INTRODUCTION 26 2 TEACHER STATISTICS 29 2.1 EMPLOYED TEACHERS 29 2.2 TEACHERS BY GENDER 35 2.3 TEACHER QUALIFICATIONS 39 2.4 TEACHER ATTRITION 44 2.5 TEACHER VACANCIES AND APPLICATIONS 46 2.6 RECOMMENDATIONS 50 3 TEACHER NEEDS AND ALLOCATION 52 3.1 MOE NORMS ON TEACHERS NEEDS 52 3.2 NEEDS ASSESSMENTS IN PRACTICE 55 3.3 TEACHER NEEDS FORECASTING 58 3.4 TEACHER ALLOCATION 61 3.5 DISCUSSION AND RECOMMENDATIONS 74 4 TEACHER RECRUITMENT 77 4.1 THE OLD TEACHER RECRUITMENT PROCESS 77 4.2 THE NEW TEACHER RECRUITMENT PROCESS 79 4.3 THE NEW TEACHER RECRUITMENT STEP-BY-STEP 80 4.4 DISCUSSION AND RECOMMENDATIONS 85 5 TEACHER REMUNERATION 93 5.1 TASHKEEL TEACHER REMUNERATION 93 5.2 NON-TASHKEEL TEACHER REMUNERATION 98 5.3 TEACHER PENSIONS 100 5.4 COMPETITIVENESS AND REAL INCOME 101 5.5 DISCUSSION AND RECOMMENDATIONS 103 6 TEACHER COSTS 106 6.1 MOE EXPENDITURES 106 6.2 TEACHER COSTS 112 Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 4 6.3 TEACHER COSTS FORECASTING 116 6.4 DISCUSSION AND RECOMMENDATIONS 120 7 REFERENCES 122 8 APPENDICES & ANNEXES 123 APPENDIX 1: OVERVIEW OF MOE CONTRACT TYPES 123 APPENDIX 2: OVERVIEW OF CADRE ALLOWANCES BY PROGRAM AND ACADEMIC TITLE 129 ANNEX 1: EDUCATIONAL INSTITUTION’S PLANNING STANDARDS 130 ANNEX 2: TWO TERMS OF REFERENCES DEVELOPED FOR NEW TEACHER POSITIONS 130 ANNEX 3: APPLICATION FORM 130 ANNEX 4: CIVIL SERVANT’S LAW 130 ANNEX 5: CIVIL SERVANT RULES AND REGULATIONS 130 ANNEX 6: PAY AND GRADES SCALE 130 ANNEX 7: BILMAKTA BILL 130 ANNEX 8: HAQUZAMA BILL 130 Photo 1: Students in class (Kabul City, 2018) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 5 Acknowledgement This report was prepared by Jain Holsheimer and Shakir Shakirullah. The team is especially thankful to the Afghanistan World Bank Education Task Team Leaders Nathalie Lahire and Matiullah (Yama) Noori for initiating this study and their advice throughout the process, as well as H.E. Dr. Mohammad Ibrahim Shinwari (Deputy Minister for General Education), Dr. Abdul Wassay Arian (Director General for Planning and Policy), Mohammad Khalid Arshad (Director General for Human Resources), Farid Mirzad (Development Budget Manager), Saleem Habibi (Payroll Manager), Najibullah Wahidi (Finance and Accounting Director), and many other officials from the Ministry of Education for their time and generous support for this study. Moreover the team also extends its gratitude to the officials from the Independent Administrative Reform and Civil Service Commission, including Dr. Hashimi (Commissioner), as well as officials from the Ministry of Higher Education, and the Central Statistics Office. The team is also indebted to Yasuhiko Matsuda (Program Leader Human Development and Governance) and the Afghanistan Governance Global Practice team, including Atiqullah Ahmadzai (Senior Public Sector Specialist) and Bernard Haven (Public Sector Specialist), for their kind support during the process as well as reviewing an early draft of this report. Moreover the team is grateful to reviewers: Tara Beteille (Senior Economist and Global Lead for Teachers), Toby Linden (Lead Education Specialist) and Yasuhiko Matsuda, for their comments and invaluable input. Furthermore, the team would like to thank Shubham Chaudhuri (World Bank country Director Afghanistan) and Abdoulaye Seck (World Bank Operations Manager Afghanistan) for their support and encouragement. Finally the team gratefully thanks the teachers, school principals, and provincial education directors who generously spent time with the team during off-duty hours. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 6 List of tables, figures, maps, and boxes Tables TABLE 1: OVERVIEW OF DIFFERENT TYPES OF MOE EMPLOYMENT CONTRACTS 30 TABLE 2: TASHKEEL TEACHERS BY GENDER 1390 AND 1396 38 TABLE 3: TASHKEEL TEACHER RECRUITS BY GENDER 1390 AND 1396 39 TABLE 4: QUALIFIED TASHKEEL TEACHERS AS PERCENTAGE OF TOTAL TASHKEEL TEACHERS 42 TABLE 5: 'ACTUAL CLASS SIZE" METHOD FOR CALCULATING TEACHER NEEDS 56 TABLE 6: 'ADJUSTED CLASS SIZE" METHOD FOR CALCULATING TEACHER NEEDS 56 TABLE 7: PUPIL-TEACHER RATIOS BETWEEN 2011-2017 63 TABLE 8: PROVINCES WITH RAPID INCREASE IN PTR 2011-2017 64 TABLE 9: COMPARISON OF PAYROLL REGISTRATION AND EMIS DATA ON TEACHERS 2017 66 TABLE 10: OVERVIEW OF DIFFERENT HAQUZAMA DATASETS AND AFMIS ESTIMATE 2017 70 TABLE 11: EXAMPLE OF HAQUZAMA POSITION DISTRIBUTION BASED ON RELATIVE SHORTAGES 73 TABLE 12: EXAMPLES OF NET SALARY CALCULATION (REGULAR AND NON-REGULAR) TEACHERS 98 Figures FIGURE 1: NUMBER OF MOE EMPLOYEE CONTRACTS PER CATEGORY 2017 31 FIGURE 2: NUMBER OF TASHKEEL TEACHERS 2011-2017 32 FIGURE 3: CONSERVATIVE ESTIMATE OF NUMBER OF HAQUZAMA CONTRACTS 33 FIGURE 4: PTR DEVELOPMENT 2011-2017 (STUDENTS OF ALL FIVE PROGRAMS) 34 FIGURE 5: MOE EMPLOYEES BY CONTRACT CATEGORY AND GENDER 2017 35 FIGURE 6: GPI DEVELOPMENT TASHKEEL TEACHERS 2011-2017 36 FIGURE 7: FORECASTING FEMALE TEACHER NEEDS TO ACHIEVE NESP III TARGETS UNTIL 2021 36 FIGURE 8: DEVELOPMENT OF TASHKEEL TEACHER QUALIFICATION 2011-2016 40 FIGURE 9: DEVELOPMENT OF TASHKEEL TEACHER QUALIFICATION LEVELS MALE/FEMALE 2011-2016 41 FIGURE 10: TASHKEEL TEACHER ATTRITION 2010-2016, IN ABSOLUTE NUMBERS 45 FIGURE 11: REASON FOR RETIREMENT AND EDUCATION LEVEL OF RETIREE 2010-2016 46 FIGURE 12: MOE STUDENT- AND TEACHER- NORMS PER CLASS 52 FIGURE 13: CONVERTED PTR NORMS AND ASSUMED TEACHING HOURS PER GRADE 53 FIGURE 14: COMPARISON OF ACTUAL REGIONAL SECONDARY EDUCATION PTRS 53 FIGURE 15: NUMBER OF ENROLLED STUDENTS 2010-2016 AND ANNUAL GROWTH IN PERCENTAGE 58 FIGURE 16: PROJECTED NUMBER OF ATTENDING AND ENROLLED STUDENTS 2017-2025 59 FIGURE 17: FORECASTED TEACHER NEEDS 2017-2025 USING ACTUAL AND ADJUSTED METHODS 60 FIGURE 18: FORECASTED TEACHER NEEDS 2017-2025 ADJUSTED TO PROGRESSIVELY ADRESS GAP 60 FIGURE 19: ANNUAL NUMBER OF ADDITIONAL TEACHER FTES NEEDED 2018-2025 61 FIGURE 20: NUMBER OF PRIMARY SCHOOLS WITHIN PTR RANGES 67 FIGURE 21: ‘STUDENT COMPENSATION BURDEN' ILLUSTRATED WITH PTR 53 IN PRIMARY EDUCATION 68 FIGURE 22: AVERAGE PTRS (ENROLLED) PRIMARY EDUCATION PER QUINTILE OF SCHOOLS 68 FIGURE 23: EXAMPLE: ‘POOLED SELECTION’ 88 FIGURE 24: EXAMPLE: ‘SINGLE NON-TRANSFERABLE APPLICATION’ 89 FIGURE 25: P&G BASE SALARY IN AFN - ALL BASTS & STEPS 93 FIGURE 26: P&G MONTHLY BASE SALARY IN AFN, REGULAR TASHKEEL TEACHERS 94 FIGURE 27: P&G MONTHLY BASE SALARY IN AFN, AJEER TASHKEEL TEACHERS 94 FIGURE 28: MONTHLY PROFESSIONAL ALLOWANCE IN AFN, BY EDUCATION QUALIFICATION 95 FIGURE 29: CADRE ALLOWANCE TTC LECTURERS, BY ACADEMIC TITLE 96 FIGURE 30: EXAMPLE OF PENSION AND TAX DEDUCTIONS OVER TEACHER GROSS SALARY 97 FIGURE 31: INSIDE- AND OUTSIDE-HAQUZAMA TEACHER SALARIES 99 FIGURE 32: SALARY RANGE OF BILMAKTA EMPLOYEES 100 FIGURE 33: CONSUMER PRICE INDEX AND REAL WAGES 2010-2016 102 FIGURE 34: OVERVIEW OF MOE EXPENDITURES PER CATEGORY 2016 106 Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 7 FIGURE 35: MOE EXPENDITURES BY CATEGORY (BILLIONS AFN) 2010-2016. 107 FIGURE 36: MOE EXPENDITURES AS PERCENTAGE OF GDP 2010-2016 108 FIGURE 37: MOE TOTAL EXPENDITURES (MILLIONS AFN) 2010-2016 AND ANNUAL EXPENDITURES 109 FIGURE 38: TOTAL MOE EXPENDITURES PER ENROLLED STUDENT 2010-2016 110 FIGURE 39: AVERAGE SALARY AND ALLOWANCES EXPENDITURE PER TASHKEEL TEACHER 2016 113 FIGURE 40: MOE BASE SALARY EXPENDITURES PER TASHKEEL TEACHER 2010-2016 114 FIGURE 41: TEACHER WAGE BILL GROWTH COMPARED TO GDP IN PERCENTAGES 117 FIGURE 42: PROJECTED MOE WAGE BILL EXPENDITURES WITHOUT INFLATION ADJUSTMENT 118 FIGURE 43: PROJECTED MOE WAGE BILL EXPENDITURES WITH FULL INFLATION ADJUSTMENT 118 FIGURE 44: ACTUAL AND PROJECTED MOE WAGE BILL EXPENDITURES AS PERCENTAGE OF GDP 119 Maps MAP 1: NUMBER OF FEMALE TASHKEEL TEACHERS PER 100 MALE TASHKEEL TEACHERS 37 MAP 2: PERCENTAGE OF QUALIFIED TEACHERS 2016 AND PERCENTAGE INCREASE 2011-2016 43 MAP 3: SHORTLISTED TEACHER APPLICANTS PER VACANCY BY PROVINCE 2018 48 MAP 4: PROVINCIAL PTRS (NUMBER OF ENROLLED STUDENTS PER TASHKEEL TEACHER) 2017 62 MAP 5: PROVINCIAL PTRS AND OVERTIME EXPENDITURE RATE, 2016 71 MAP 6: PROVINCIAL NET PTRS (BASED ON AFMIS ESTIMATE) 2017 72 MAP 7: TOTAL MOE ANNUAL EXPENDITURES PER ENROLLED STUDENT BY PROVINCE, 111 MAP 8: WAGES AND SALARIES ANNUAL EXPENDITURES PER ENROLLED STUDENT BY PROVINCE, 112 MAP 9: BASE SALARY EXPENDITURE PER TASHKEEL TEACHER (AVERAGE 2010-2016) 115 Boxes BOX 1: HAQUZAMA CONTRACTS, THE ‘HIDDEN’ TEACHER WORKFORCE 32 BOX 2: TEACHER DATABASES, DATA GAPS, AND DATA INTEGRATION 49 Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 8 Acronyms & Glossary Acronyms AFMIS Afghanistan Financial Management MEC Joint Anti-Corruption Monitoring Information System and Evaluation Committee AFN Afghan Afghani (Currency) MoE Ministry of Education BPHS Basic Package of Health Services MoHE Ministry of Higher Education CSO Central Statistics Office NESP National Education Strategic Plan CBE Community-Based Education N/A Not Applicable/Available CPI Consumer Price Index NGO Non-Governmental Organization EMIS Education Management Information OMR Optical Mark Readers System DED District Education Directorate O&M Operations & Maintenance EFA Education For All PED Provincial Education Directorate EPHS Essential Package of Hospital Services PhD Doctor of Philosophy FTE Full-Time Equivalent PTR Pupil-Teacher Ratio GDHR General Directorate for Human P&G Pay and Grade Resources GDP Gross Domestic Product QBD Question Bank Database GD P&P General Directorate for Planning and SDED Student Data Entry Database Policy GE General Education TED Teacher Education Directorate GNP Gross National Product ToR Terms of Reference GPI Gender Parity Index TTC Teacher Training College HR Human Resources TVET Technical Vocational Training and Education IARCSC Independent Administrative Reform UNESCO United Nations Educational, and Civil Service Commission Scientific and Cultural Organization IDP Internally Displace People UNICEF United Nations Children’s Fund ID Identification Document USD United States Dollar IE Islamic Education Y-Y Year on Year IMF International Monetary Fund Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 9 Glossary Afghani Afghanistan’s national currency Ajeer Support staff, typically cleaners, guards, drivers, tea makers, etc. Bast A civil servant’s grade or rank (based on his/her position); Bast 1 being the highest rank and Bast 8 being the lowest rank. Bilmakta Temporary contract modality for public sector employees/positions Haquzama Temporary contract modality used specifically for academic employees/positions Ikramya A lump sum amount paid out after retirement or death, similar to a ‘golden handshake’. Islamiat Islamic Religious Studies Kadam A civil servant’s grade or rank (based on work experience); Kadam 1 being the lowest and Kadam 5 being the highest. The report uses the term STEP instead of Kadam. Kankor Derived from the French ‘concourse’ meaning competitive; typically refers to the competitive university entry exam. Tashkeel The Tashkeel (noun) refers to the approved (civil service) staffing structure; tashkeel (adjective) refers to a position or contract within the approved staffing structure. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 10 Executive Summary 1. This study aims to present a comprehensive analysis of teacher management issues in Afghanistan, in order to inform the Ministry of Education’s formulation of a new teacher management policy, and to build consensus amongst education development partners on priorities for reform and assistance in the short (within the next 12 months) and medium (within the next 36 to 48 months) terms. This study uses a historical approach to teacher management issues (planning, HR, financial management), describing past developments to inform future priorities. Since 2002, the Ministry of Education (MoE) has made tremendous progress in its capacity to manage the teacher workforce, despite the multitude of challenges it faced. However, not all of Afghanistan’s provinces have witnessed improvements and growth evenly. In part, this report describes how progress has been uneven, and calls for a more diversified teacher management approach that considers regional, gender and urban vs. rural discrepancies, in order to promote equity, efficiency and effectiveness. 2. This report includes analyses of Teacher Statistics (Chapter 2), including their numbers, gender and qualifications; Teacher Needs and Allocation (Chapter 3) describes how the MoE determines teacher needs, and how teachers are distributed across provinces and schools; Teacher Recruitment (Chapter 4) describes how teachers were recruited in the past and assesses the recently introduced reforms in teacher recruitment; Teacher Remuneration (Chapter 5) describes teacher remuneration, including base salaries, allowances, deductions and pensions; and finally, Teacher Costs (Chapter 6) analyses per student expenditures and forecasts teacher costs. 3. This study combined quantitative data analyses of existing datasets with qualitative interviews. Datasets were retrieved, and interviews were conducted between October 2017 and March 2018. This study focused exclusive on teachers directly managed, administered and paid by the Ministry of Education, and does not cover teachers working in community-based education (CBE), private education, or NGO-provided education. Teacher Statistics 4. There is no accurate, centralized, and up-to-date registry of all teachers employed by the MoE. There are multiple databases on teachers, but not all are frequently updated, and none of them cover all types of teacher contracts. The most comprehensive and accurate database on employed teachers is the payroll registration database. However, not all employees are correctly labeled as teachers or non-teachers in this database, and it does not include all types of temporary (haquzama) contract teachers. Different datasets on haquzama teachers showed conflicting numbers. As a result, this study was unable to determine the number of people employed as teachers by the MoE, or the number of teacher full time equivalents (FTEs), with full accuracy. 5. In 2017, there were approximately 218,261 people employed as teachers by the MoE. This includes 203,641 tashkeel teachers (teachers employed in the approved ‘permanent’ staffing structure), and an additional 14,620 temporarily contracted teachers (haquzama outside tashkeel). However, a significant number of tashkeel teachers have an additional second full-time teacher contract (haquzama inside tashkeel), and the total number of teacher FTEs is estimated at 239,465. The number of tashkeel teachers increased significantly prior to 2013 and thereafter plateaued. Data on haquzama contracts appeared unreliable and exact numbers Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 11 could not be determined. This posed a challenge to almost all analyses of teacher statistics in this report. 6. Between 2011 and 2017, the number of enrolled students increased faster than the number of tashkeel teachers and haquzama contracts. As a result, the national pupil-teacher ratio (PTR), dividing the number of enrolled students by the number of tashkeel teachers, rose from 41 in 2011, to 46 in 2017. However, if the PTR is calculated using attending students and total teacher FTEs, the index increased from 28 in 2011, to 30 in 2017. 7. 33% of tashkeel teachers are female and 67% are male (2017). Converted, the gender parity index (GPI: the number of females per male) for tashkeel teachers is 0.50. In the past seven years, the GPI for tashkeel teachers has improved somewhat. In 2011 there were only 44 female tashkeel teachers per 100 male teachers (GPI 0.44), in 2017 this number had risen to 50. However, at the current pace the MoE is not reaching its targets of GPI 0.69 by 2021, as expressed in the National Education Strategic Plan (NESP III). Given limited turnover in the teacher workforce, the MoE has limited opportunities in improving the GPI, and addressing the gender-gap will require far reaching policy reforms in teacher recruitment. 8. More than half of all female tashkeel teachers in Afghanistan work in Kabul city, Balkh or Herat province, and the national average GPI hides extreme differences in the teacher gender gap between provinces. In the 17 provinces with the lowest GPI, there are on average only 12 female teachers per 100 male teachers (GPI 0.12). Importantly, in the past seven years, those provinces with the least female teachers (in 2011) have made almost no progress in recruiting more female teachers. In these provinces, improvements in the GPI are marginal (+ 0.02), whereas in the provinces with the highest GPI (in 2011), progress is more significant (+ 0.11). 9. The minimum education qualification of a ‘qualified’ tashkeel teacher (teaching grades 1 to 12 in primary, lower and higher secondary) is ‘Grade 14’, the equivalent of high school plus two additional years (usually completed at a teacher training college). The number of ‘qualified’ tashkeel teachers almost doubled in recent years, from 61,091 in 2011, to 114,762 in 2016. And the percentage of tashkeel teachers with a grade 14 (or higher) qualification rose from 33% in 2011, to 57% in 2016. Two-thirds of female teachers (66%) are qualified, whereas 53% of male teachers are qualified (in 2016). However, the national averages hide dramatic differences in the qualifications of teachers across different provinces. In some provinces, including Paktika, Daikundi, Zabul, Kandahar, Uruzgan, and Badghis, less than 20% of teachers are ‘grade 14’ educated. In Kabul City, Balkh, Jawzjan, Faryab and Kapisa, more than 75% of teachers are ‘grade 14’ educated (in 2016). Qualification discrepancies across provinces have not decreased in the past seven years. 10. The teacher qualification gap compounds the teacher gender gap (r=.47). The provinces with the lowest percentage of female teachers, also tend to have the lowest percentage of ‘grade 14’ female teachers. In Wardak, Paktya, Zabul, Uruzgan, Paktika and Khost province, less than 1% of all tashkeel teachers are ‘qualified’ female teachers, and these provinces made almost no progress improving female teacher qualifications between 2011 and 2016. 11. Provincial teacher qualification gaps are further amplified by discrepancies in subject qualifications. Provinces with few qualified teachers (Paktika, Daikundi, Zabul, Kandahar, Uruzgan and Badghis) have even fewer subject qualified teachers that studied the subject they teach. Intra-provincial differences in subject qualifications are more pronounced. Within Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 12 provinces with very few qualified teachers, it is typically the Islamiat teachers that are subject qualified and for other subjects there are almost no subject qualified teachers. 12. There is no nation-wide standardized testing of teacher abilities within the pool of MoE employed teachers. However, a recent study1 suggests that a significant portion of the teachers is severely under qualified. For example, half of the (primary school) teachers tested, were unable to divide double digits by single digits or complete a number sequence. 13. Teacher attrition rates are low; in the 2010-2016 period between 0.6% and 1.0% of tashkeel teachers retired from active duty annually. That is, each year, out of a thousand tashkeel teachers, between six and ten teachers retired, on average 1,570 teachers per year. This category includes termination of employment due to retirement, death and voluntary contract termination. The attrition rates for female teachers are somewhat lower than for male teachers. However, datasets on teacher attrition appear to be incomplete or inaccurate. 14. There is no centralized MoE database on teacher vacancies or applications. This study was not able to find a list of teacher vacancies or applications across provinces. Comparing different datasets, this study estimated 11,525 teacher vacancies across the country in 2017, or on average 329 tashkeel teacher vacancies per province. Estimated vacancies are evenly distributed across most provinces (between 1% and 10% vacant positions). Four provinces have an exceptionally high percentage of vacancies, including Zabul (51%), Daikundi (49%), Panjshir (24%) and Badghis (19%). However, dataset discrepancies are likely inflating estimates in this calculation. 15. At the time of writing (February/March 2018), the Independent Administrative Reform and Civil Service Commission (IARCSC) was in the process of hiring civil servants through a ‘new’ reformed recruitment process. The IARCSC shortlist included 118,544 applicants for 8,000 teacher vacancies nation-wide. The number of shortlisted applicants varies considerably across provinces. In some provinces, on average more than forty applicants were shortlisted for each available teacher position in that province (51 in Faryab, 45 in Baghlan), suggesting significant interest in the teacher profession there, as well as an adequate supply of eligible candidates. Conversely, in some provinces, on average less than two applicants were shortlisted for each available teacher position in that province (1 in Uruzgan, and 1 in Paktika), suggesting a significant shortage of willing and eligible candidates for teacher positions. Importantly, not all shortlisted applicants completed registration, and as a result, there were less registered teacher applicants in Uruzgan than teacher vacancies. Teacher Needs 16. The Ministry of Education norms used to calculate the number of teachers needed for schools are described in a policy document called: Educational Institution’s Planning Standards (2007). The planning standards use a class-based method to identify teacher needs per grade. The MoE’s norms for secondary education, between 20 and 23 students per teacher, are comparable to international trends, both regionally and compared to other low-income countries. 17. There are four main issues that limit the practical application of these planning standards for planning purposes. First, the existing norms on staffing schools are aspirational rather than realistic. If fully applied, the MoE would have to double its workforce. Second, the 1The Learning Crisis in Afghanistan, Results of the SABER Service Delivery Survey 2017 (PRELIMINARY DRAFT). P.56-63. The World Bank, 2018. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 13 norms are not comprehensive, not detailed, and not consistent. Third, the norms do not identify a method to prioritize teachers (and school staff) needs under conditions of scarcity. And, last, MoE staff involved in the assessment of teacher needs do not accurately know the needs assessment norms. As a result, the norms applied to identify teacher needs vary considerably. 18. The MoE uses two different processes to assess teacher needs. The first is an internal consultative process where the central Ministry asks its provincial education directorates (PEDs) to provide overviews of teachers needed (supported by district education directorates and schools). The reliability of this process is problematic, as there seem to be a multitude of different norms and calculation methods applied by officials assessing teacher needs. As a result, two schools with the same number of students, classes and teachers, calculate different numbers of additional teachers needed. In the second process, the MoE general directorate of planning and policy (GD P&P) uses its database on students, classes and schools (the Education Management Information System, or EMIS) and teachers (the payroll registration database) to determine the number of teachers needed across the country. However, methods for calculating teacher needs at the central MoE are also not standardized. 19. This reports presents two methods used by the MoE GD P&P to assess teacher needs. The first method (i.e. the ‘adjusted class size’ method) calculates teacher needs based on the number of students, applying the norms as averages (e.g. there should be on average 1 teacher per 20 students). The second method (i.e. the ‘actual class size’ method) calculates teacher needs based on the number of classes and applies norms as upper limits of class size (i.e. there should be between 10 and 20 students per class). The methods have significantly different outcomes. The first method concludes 26,268 additional teachers are required (in 2017); the second concludes 63,048 additional teachers are required (both calculations exclude assistant teachers and support staff). However, none of the methods used by the GD P&P considers the numbers of haquzama teachers, and the estimated number of teacher FTEs (in 2017) is comparable to the number of teachers needed using the first method. 20. Given variations in class size, the use of the second ‘actual class size’ method is sensible, but also problematic, as it means that teacher needs are effectively determined by class size. The more classes with few students there are, the more teachers are needed, even with the same amount of total students. The policy of allowing the creation of classes at a minimum of ten enrolled students, in order to promote access to education in rural areas, has resulted in a very large number of schools with class sizes lower than the norm. Moreover, the ‘actual class size’ method to calculate teacher needs, requires detailed knowledge of the size of each class. The MoE has also not identified average target PTRs per province or nation-wide, and neither has it published its methods for calculating teacher needs. This prevents national, subnational, and international education stakeholders from externally validating the teacher needs assessments. 21. Teacher needs forecasting is challenging because it is difficult to predict future student numbers without reliable indicators. EMIS data shows an increase of 30.3% in enrolled students between 2010 and 2017 (approximately 3.8% year-on-year). Teacher needs until 2025 are forecasted using the two different calculation methods, and assuming student enrollment trends will be similar to the past seven years. However, an estimated 3.5 million children were out of school in 2017 (girls comprise 75% of this group)2 and if the MoE were to effectively address this issue, student enrollment, as well as teachers needed, would rise more rapidly. 2 UNICEF Afghanistan Annual Report 2017, p3. UNICEF 2017. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 14 22. If student enrollment continues to increase as it did in the past seven years, the MoE will need around ten thousand additional teacher FTEs each year, only to maintain the current PTR. Using the first ‘adjusted class size’ method an additional 9,274 teachers will be required in 2018, and this number will progressively increase to 12,100 additional teachers in 2025. The total number of additional teacher FTEs needs, will increase by 85,071 (36%) until 2025. Using the second ‘actual class size’ method, an additional 13,920 teacher will be required in 2018, and this number will progressively increase to 20,673 additional teachers in 2025. The total number of additional teacher FTEs needs, will increase by 136,844 (57%) until 2025. 23. However, the impact of additional teachers will be severely constrained if the MoE does not improve the efficiency of its teacher allocation and distribution methods. Current pupil- teacher ratios are extremely uneven across provinces, districts and schools, resulting in the inefficient use of the teacher workforce. Furthermore, the lack of accurate teacher data, and the lack of transparent methods for calculating teacher needs, prevent a consensus on what are fiscally sustainable teacher norms. As such, a significant year-on-year increase in the number of teachers will be meaningless without effective reform in teacher planning. Teacher Allocation 24. In 2017, some provinces had twice as many tashkeel teachers as other provinces, for the same number of students. Nuristan, Panjshir, Badakhshan, Wardak, Parwan, and Baghlan each have less than 36 enrolled students per tashkeel teacher, whereas Helmand, Kandahar and Khost have between 60 and 72 students per teacher. The significant differences in these PTRs cannot be explained by either the geographic composition of these provinces (i.e. more small classes in rural schools), or the relative size of grades (i.e. the age composition of students). There are justifiable reasons to differentiate provincial PTR targets/norms. For example, less urbanized provinces might need smaller schools (with lower PTRs) to provide access to rural communities. However, the MoE does not have any written policy on PTR norms, nor on differentiating teacher needs based on degree of ‘urbanization’/concentration. 25. In the past seven years, the significant variations in provincial PTRs have not been decreased through the allocation of new tashkeel teacher positions. Moreover, in several provinces the number of students rose dramatically between 2010 and 2016, but these provinces did not receive significantly more teacher positions than the national average. In late 2017, the MoE allocated 8,000 new tashkeel teacher positions across its 35 PEDs. All provinces received additional teacher positions, including those that already had very low PTRs (Parwan, Wardak, Panjshir, Badakhshan and Baghlan), and the provincial PTR gap has not been reduced by this distribution. Most MoE officials interviewed for this study did not know how provincial teacher allocations were determined, or which department was responsible for this. 26. Data on the distribution of teachers across schools is less reliable. EMIS is based on self- reported data by schools. There are tremendous variations in EMIS reported PTRs across schools. Almost 23% of primary schools (n=1,295) reported having more than 100 students per teacher, and almost 13% (n=719) reported having more than 140. Half of the schools reported having between 20 and 70 students per teacher, and 11% (n=620) reported having less than 20 students per teacher. These variations might appear extreme, but are not unexpected, given the actual PTR in primary education (53 on average) and the policy to allow the creation of small classes. While more reliable (and verified) data is required to be conclusive, it is clear that there are substantial discrepancies between schools. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 15 27. Unlike tashkeel positions, the numbers of haquzama teachers (and their provincial distribution) are not centrally determined or approved. The primary limitation on haquzama contracts is budget availability. AFMIS ‘overtime’ expenditure data provides an estimate of the relative distribution of haquzama teachers (haquzama salaries are paid out of the ‘overtime’ budget line). Overtime expenditure rates, as a percentage of teacher salary expenditures, vary significantly across provinces. Interviews and data analyses suggest that there is no clear policy or regulation on the number of haquzama contracts and how they are prioritized across schools and provinces. 28. Overtime expenditures (as an estimate of haquzama teachers) are not prioritized across provinces with high PTRs. Most notably, Panjshir province spent 118 million AFN (1.7 million USD) on teacher salaries in 2016, and an additional 63 million AFN (0.9 million USD) on overtime, even though Panjshir already had the lowest PTR in the country (based on enrolled students per tashkeel teacher). As a result, the estimated NET PTR (attending students per teacher FTE) in Panjshir, is as low as 14, whereas this is 42 in Helmand and 38 in Khost. 29. Provincial NET PTRs can be equalized (almost fully) by only redistributing temporary haquzama contracts based on the relative shortage in each province. To balance the number of attending students per teacher FTE across provinces, it is not necessary to terminate or relocate tashkeel teachers. Only seven PEDs (Parwan, Wardak, Badakhshan, Baghlan, Bamiyan, Nuristan and Panjshir) currently have more tashkeel teachers than required based on the national average of attending students (29.8) per teacher FTE. Provincial NET PTRs can be fully equalized after 5 years, by prioritizing new tashkeel position only to those provinces with current shortages. 30. Current methods used to assess teacher needs, and distribute them across provinces and schools, are not standardized, and institutionalized, nor are they transparent, or replicable. This thwarts both a political and technical debate about the necessity of more teachers. Furthermore, current methods result in significant discrepancies between provinces and extreme discrepancies between schools. This is neither fair nor efficient. Afghan children sharing a teacher with one hundred other students do net get the same learning opportunities as children sharing their teacher with only five or ten other students. 31. The MoE needs to critically evaluate the costs of its minimum benchmark on creating classes. The effective minimum benchmark for creating a ‘grade 12’ class is 5 attending students per teacher. The MoE needs to consider the benefits of this policy against the costs of having enormous classes (100+). In addition, the MoE’s existing flexibility in human resource allocation (i.e. the temporary haquzama teacher contracts) is not optimally used to address Afghanistan’s volatility in education needs. The data analyses suggests that the MoE can significantly improve both fairness and efficiency in the public education sector, by adopting norms and methods that result in a more equitable teacher distribution. Teacher Recruitment 32. Effective and merit-based teacher recruitment is a critical tool for promoting teacher quality. However, in the past years, the merit-based recruitment of new teachers appears to have been significantly undermined by nepotism and corruption 3 . Until 2018, teacher recruitment was managed and implemented at the provincial level. PEDs openly deviated from 3Ministry-wide Vulnerability to Corruption Assessment of the Ministry of Education. Independent Joint Anti- Corruption Monitoring and Evaluation Committee (October 2017), p2. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 16 national civil service regulations4, and the recruitment process appeared to be vulnerable to corruption at almost every step. Already in 2015, an Independent Joint Anti-Corruption Monitoring and Evaluation Committee (MEC) report5 stressed the vulnerabilities to corruption across all steps of the teacher recruitment process. Many MoE officials at national, provincial and school level, interviewed for this study, expressed their believes that abuse of teacher recruitment was widespread. As a result of decentralized implementation, there is also no quantitative data available in Kabul on new teacher recruits (i.e. the number of new recruits and their qualifications). Furthermore, there is no information available on the quality of the recruitment process (fairness, transparency, merit-based, etc.). 33. In 2017, significant reforms in the teacher recruitment process were implemented. This ‘new’ teacher recruitment process differed in significant ways from past recruitment practices. First, the IARCSC became the lead implementing agency for the new teacher recruitment, with a significant role for the Ministry of Higher Education (MoHE) to support this. Second, some key parts of the recruitment process (registration, exam and questionnaire preparation, scoring, announcement) are now centralized in Kabul. And third, the teacher entry examination process is implemented in the same way, and governed by the same rules, as the kankor university entry exam. At the time of writing, the IARCSC had just finalized the application and shortlisting process and the first teacher entry exams took place in the East and South of the country. This study was unable to fully and accurately evaluate the effectiveness of the reforms, due to ongoing implementation and inherent limitations of this study. A full and complete evaluation of the new recruitment process would enable a more accurate assessment of how effective these reforms have been. 34. The new recruitment process addresses some of the most critical corruption vulnerabilities identified in the old process. First, the creation of a centralized electronic database on applicants that registers all steps of an individual in the recruitment process (application, examination, selection) is a tremendous step forward in addressing previous corruption vulnerabilities. Second, the new entry examination method almost completely eliminates opportunities of abuse. Third, the unique and randomized exam questionnaire and answer sheets allow for a gradual (province by province) and flexible roll out of a nation-wide civil servant recruitment effort. Fourth, the ‘single non-transferable application’ selection method (a group of applicants only compete for a single vacancy) eliminates opportunities for favoritism in the final selection and appointment stage. And last, the central electronic database on teacher applicants, combined with exam scores, provides tremendous information to further improve teacher policies. 35. While it is arguably too early to discern the impact of the reforms, this study identified areas for improvement in future recruitment rounds. First, the new teacher recruitment process lacked a clear description, regulation and accountability mechanism, and the IARCSC has not provided adequate instructions to applicants on how the application, exam and selection processes work. From the perspective of the applicant, it is difficult to verify the exam score or the fairness of the selection process. While the new recruitment process seems to have significantly reduced vulnerabilities to actual corruption, it has done little to address the wide spread perception of corruption. 4Education Sub-National Assessment: Strengthening Public Education Management. Altai Consulting. (2015), p.62. 5Vulnerability to Corruption Assessment of Teacher Recruitment in the Ministry of Education. Independent Joint Anti- Corruption Monitoring and Evaluation Committee (June 2015), p7-12. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 17 36. Second, it is yet unclear to what extend the IARCSC was successful in promoting female and rural teacher recruitment. Targeted female recruitment is the most effective tool to increase women into the teacher workforce. According to the IARCSC, female candidates for the teacher entry exam received additional bonus points increasing their chances of being selected. In addition, teacher vacancies were gender specific. However, it is yet unclear to what extend the IARCSC was successful in increasing the number of women hired into the teacher workforce, especially in those provinces where the gender gap is most pronounced. The reforms also have done little to facilitate rural teacher applicants. Although too early to assess at this point, the initial data published by the IARCSC shows that in some rural provinces (e.g. Uruzgan) only half of the shortlisted candidates collected their exam ID and submitted their biometric data. And last, the selection method applied in the new process, will potentially result in a significant loss of qualified and willing candidates, as candidates apply only for a single vacancy, and qualified but unsuccessful candidates are not offered alternative positions. Teacher Remuneration 37. Tashkeel teacher remuneration (i.e. a ‘regular teacher salary’) is regulated by national civil service laws, regulations, and scales. These rules and scales are the same everywhere in Afghanistan. There are no regional differences in salaries or allowances, nor are there formal institutionalized differences between men and women or urban and rural teachers. ‘Base’ salaries of all civil servants (including teachers) are determined by the civil servant pay & grade (P&G) salary scale. This scale, divides civil servant positions into eight grades or ‘basts’. Bast 1 is the highest grade, reserved for senior civil servant positions (director general level), and bast 8 is the lowest grade, reserved for support staff (e.g. cleaners, guards, etc.). Basts are subdivided into five STEPS reflecting work experience. 38. P&G base salaries of regular (qualified) tashkeel teachers (bast 5 and bast 6) range from 6,500 AFN to 9,200 ANF per month (93 USD to 131 USD). However, for some teachers (under qualified teachers and teachers that have not been graded according to P&G yet) the base salary is lower or higher than this range. In addition to the P&G base salary, civil servants receive a professional allowance dependent on their level of education. This allowance aims to incentivize the recruitment and retention of educated professionals and ranges from 8 AFN per month for ‘grade 12’ educated teachers, to 6,000 AFN for teachers with a PhD (0.11 USD to 86 USD). Lecturers that teach tertiary education (classes above grade 12), receive an additional cadre allowance. Lecturers work at TTCs, TVET institutes, and some literacy and Islamic Education institutes. The allowance is determined by the academic title and ranges from 8,000 AFN to 12,000 AFN (114 USD to 171 USD). All civil servants, and all tashkeel teachers receive an additional food allowance. This food allowance is a fixed 30 AFN (0.43 USD) per day, or 660 AFN (9.41 USD) monthly. Tashkeel teachers do not receive a transportation allowance, although regional exceptions to this rule exist. 39. Tashkeel teachers pay an eight percent pension contribution deducted from their gross salaries. This is the most significant deduction of a typical gross teacher salary. Furthermore, teachers typically pay around 1% tax (effective tax rate) on their salary and allowances. Teacher salaries are deducted if a teacher exceeds the maximum number of leave days. Tashkeel teachers are expected to work nine months per year. Within those nine months, they are allowed to take up to 30 leave days. If a teacher exceeds his or her leave limits, both the base salary and professional and cadre allowances are deducted. 40. Salaries of non-tashkeel teacher contracts are regulated nationally through specific bills. The salaries of haquzama contracts are calculated based on a fixed hourly fee, depending on Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 18 the education level of the contracted teacher. Haquzama contracts for general education teachers (in primary and secondary education) are always for 24 hours per week, and thus one teacher FTE. Monthly salaries range from 5,760 AFN (82 USD) for grade 14 educated contract teachers; 6,720 AFN (96 USD) for grade 16 educated; and 7,680 AFN (109 USD) for haquzama teachers with a master’s degree. Haquzama contract teachers do not receive professional or cadre allowances, nor do they receive lunch or transportation allowances. Haquzama salaries are deducted for tax but not for pensions. Salaries for bilmakta-ajeer teacher contracts (the least common contract within the pool of teachers) are not fixed. The bilmakta regulation indicates that salaries need to be negotiated within a range from 5,000 AFN (71 USD) to 13,000 AFN (185 USD). 41. Teacher pensions are regulated by the civil servants law and regulations, and pensions are the same for teachers as well as other civil servants. There are different conditions under which a teacher can retire and/or receive a pension (long serving, old age, early retirement, death, or voluntary termination). The size of pension payment depends on the duration of service and the size of the last monthly salary. In addition, under certain circumstances, retiree receives ‘Ikramya’ (a golden handshake). The retiree’s golden handshake is ten times the last monthly salary paid at once. Last, the pension is paid for the remainder of the retiree’s life. 42. Public teacher salaries are comparable to private sector remuneration. Private school teacher salaries are generally not perceived as better than public school teacher remuneration. While some teachers in the private sector earn more than their public sector colleagues, most earn the same or less. However, current salaries of public school teachers, working 24 to 28 hours per week, are not perceived as sufficient to sustain a family. For male teachers, a second job (e.g. as a shopkeeper, or private tutor) is the rule rather than the exception, according to teachers interviewed for this study. In 2015 (1394), teachers in the capital Kabul protested the rising cost of living and the perceived lack of adequate remuneration. 43. Since the introduction of the P&G scale in 2008, civil servant (and teacher) salaries have not been adjusted for inflation, despite significant year-on-year inflation in the same period. Some MoE officials interviewed for this study stressed that regular civil servants, rather than teachers, suffered the most from the depreciation of their (real) income, since they are expected to work a 40-hour workweek and are less able to generate income from a secondary job. However, the latest round of teacher recruitment in early 2018, suggests that the teacher profession remains attractive for job seekers (118,544 applications were shortlisted for teacher vacancies). 44. The current remuneration scale and regulations do not enable incentivizing policy priorities. Remuneration scales are equal across provinces and districts (although regional expenditures per teacher vary), and the same for men and women, rural and urban teachers, even though there are considerable regional and gender variations. Rural schools across Afghanistan are struggling to find qualified candidates for their positions, and qualified female candidates are exceptionally hard to find outside of urban centers. Yet, at the moment, a female teacher in Deh Rawud, rural Uruzgan, receives the same remuneration package as a male teacher in Kabul city, and the MoE has no tools to incentivize female, regional or rural recruitment. Other public sectors (e.g. public health) do allow for regional, gender and hazard allowances. But the Ministry of Education lacks the necessary remuneration tools to provide equitable education services to all children in Afghanistan. 45. Teacher remuneration is not adapted to the unique characteristics of the teacher profession. Teachers do not change the ‘type of position’ during the course of their careers, Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 19 and thus they cannot promote across basts. Teacher salaries change only the first four years of a teacher’s career, after which the salary remains unchanged for the remainder his/her teacher career. Teachers also work significantly fewer hours per week than regular civil servants (24 to 28 hours per week)6. Furthermore, unlike regular civil servants, considerable numbers of teachers work in rural areas (i.e. below district capital level), and there is a higher demand for female employees in the teacher workforce to promote access to education for girls. The unique features and requirements of the teacher profession, combined with the size of the workforce (more than half of all civil servants excluding security services) warrant reforms to enable more effective teacher management. Currently, only education qualification is incentivized through the professional allowances. Teacher Costs 46. Teacher base salaries alone are responsible for half of all MoE expenditures. In 2016, the MoE spent almost 18 billion AFN (256 million USD) on teacher base salaries. In addition the MoE spent 4.3 Billion AFN (61 million USD) on non-teacher base salaries, 3.6 billion AFN (52 million USD) on ‘overtime’ payments, and 2.4 billion (34 million USD) on other wage allowances (including food, professional and cadre allowances). The MoE wage bill absorbs 78% of all MoE expenditures. MoE total expenditures increased by 61% from 2010 to 2014, and thereafter plateaued. 47. Although education expenditures increased (2010-2016), MoE expenditures compared to Afghanistan’s nominal GDP declined in the 2010-2016 period, from 3.17% to 2.79%. The relative distribution of MoE expenditures per category did not fluctuate significantly in past seven years, except ‘overtime’ expenditures. MoE overtime expenditures rose by 260% between 2010 and 2016, from 1 billion AFN (15 million USD) to 3.6 billion AFN (52 million USD). The magnitude of MoE’s expenditures on overtime is significant. Actual 2016 MoE expenditures on ‘overtime’ would be sufficient to cover between 51,800 and 60,500 additional haquzama FTEs (although overtime payments of non-teacher MoE staff are also paid out of this budget). Moreover, it suggests that the MoE datasets on haquzama contract are likely underestimating the actual number of haquzama FTEs. 48. In 2016, the MoE spent on average 4,050 AFN (58 USD) per enrolled student. This includes wages and salaries, overtime, O&M and development expenditures. Expenditure per enrolled student increased by 33% between 2010 and 2014, and thereafter decreased somewhat. Average annual expenditures per student in the past seven years, was almost three times higher in Panjshir than in Badghis. In 2016, the MoE spent on average 3,179 AFN (45 USD) on wages and salaries per enrolled student. Wages and salary expenditures per student vary significantly from 1,804 AFN (26 USD) in Uruzgan, to 6,317 AFN (90 USD) in Panjshir. 49. This study analyzed seven-year averages on per student expenditure, to correct for year- on-year fluctuations in provincial expenditure data. Still, the MoE spent three times more on wages and salaries per student in Panjshir than in Badghis, on average between 2010-2016. In addition to PTRs, there are two factors determining per student expenditures on wages and salaries. Provinces with low wage expenditures per student (Badghis, Paktika, Daikundi, 6Importantly, this study was unable to find an official policy or guideline or teacher work hours at the MoE. MoE officials involved in provincial level HR management of teachers were confused and disagreed about the number of officials working hours. Teachers interviewed for this study indicated that most teachers work between 24 and 28 hours a week. While this is comparable to international standards, teachers indicated that work before and after class was not the norm. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 20 Uruzgan) typically have a combination of low overtime expenditure rates and low average teacher costs. Provinces with high wages and salary expenditures per students typically have high rates of overtime expenditures and high average teacher costs, in addition to low PTRs. 50. In 2016, the MoE spent 7,455 AFN (106 USD) on teacher base salaries per tashkeel teacher per month, and 89,457 AFN (1,278 USD) per year. In addition, the MoE spent on average 645 AFN per tashkeel teacher on food allowances, and 281 AFN on professional and cadre allowances, putting average monthly expenditures (base salary plus allowances) per tashkeel teacher at 8,381 AFN, and annually at 100,564 AFN (120 USD and 1,437 USD). 51. Average (base) salary expenditures per tashkeel teacher rose by 42% between 2010 and 2016. Teacher intra- and inter-bast promotion can only partially explain this increase. An additional likely explanation is the gradual implementation of pay and grade (P&G) reform. Many teacher salaries were not adjusted according to the current P&G salary scale in 2010. Adjusting teacher salaries in the 2010-2016 period might have caused the significant increase in base salary expenditures. More rigorous analysis is needed to understand the 42% increase in per tashkeel teacher base salary expenditures. 52. Significant regional differences exist between average base salary expenditure per tashkeel teacher. In 2016, the Kabul City PED spent on average 8,650 AFN (124 USD) per registered tashkeel teacher per month, whereas the Paktika PED spent 4,647 AFN (66 USD) per month per teacher. Average base salary expenditure correlates with the percentage of qualified teachers in a province(r= .65). However, there are notable outliers from this trend. 53. Another method of calculating teacher costs is to divide the total MoE Wages and Salaries expenditures (AFMIS code 21) by the number of MoE employed teachers (in FTEs). This indicator more accurate reflects costs per teacher and is used as the primary driver to forecast future MoE wage bill costs. 54. Future teachers costs are determined by average teacher salaries (which increase due to teacher promotion and limited turnover), salary inflation adjustments, and the number of teachers. Average salaries will continue to increase in the coming seven years, due to the increase in average bast and STEP of teachers, and the corresponding incremental increase in salaries determined by the P&G pay scale. Salary inflation adjustments (this study did not conclusive analyses pay competitiveness and does not propose indexing salaries to inflation), be it partially, will be necessary to compensate for real wage loss, and the teacher workforce will need to continue to grow year-on-year to maintain current PTRs. 55. Maintaining current PTRs (3.8% y-y teacher growth), combined with anticipated average salary increases due to promotion (2.1%) and partial inflation adjustments will not significantly increase education expenditures compared to anticipated GDP or overall government expenditures (based on IMF projections). Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 21 Recommendations – Teacher Statistics The Ministry of Education should: SHORT TERM: (within the next 12 months) 56. Immediately clarify existing ambiguities regarding temporary (haquzama) teachers, including numbers and distributions across provinces, districts and schools. Information on haquzama contracts should be reported systematically and periodically verified. Haquzama data should include gender, age, qualifications and school (work location) similar to information held in regular tashkeel employee profiles. Temporary (haquzama) teachers need be part of any report or presentation by the Ministry on its workforce. Data collection on haquzama teachers should also include specification on the contract’s FTEs. Clarify Ministry employee categories. Separate teaching and non-teaching employees into different categories; identify hours worked per day/week/month and hours in front of class. Create separate categories for school support staff and MoE/PED/DED support staff. 57. Develop a clear strategy to improve the teacher GPI. Identify feasible annual targets nationally and for each province. Assess the key impediments to female recruitment in provinces lagging behind in GPI improvements, and those where improvements have been only marginal in the past years. Identify which tools can be used to promote female recruitment, and develop a comprehensive policy that enables to use of these tools in target provinces. 58. Periodically apply standardized in-service tests to all teachers, and identify teachers, subject qualifications, and regions that should be prioritized for in-service training. Standardized tests (preferably applied to all teachers periodically) will help identify key priority areas and subjects that require a focus for in-service teacher training. Address the issue of severely under qualified teachers. Identify minimum qualification standards for in-service teachers (e.g. reading, writing, basic mathematically equations), test all in-service teachers, and terminate tashkeel contracts of those in-service teachers that do not meet the minimum standards. An appropriate policy needs to be developed to suggest termination, in-service training, or continuation based on test results, and to guarantee rigorous oversight of standardized in-service tests. Furthermore, minimum qualification standards should be specified for teachers in different grades and school types (primary, lower and higher secondary). 59. Develop a comprehensive, multi-year plan to address pre-service teacher training needs by province, district, gender, and subject specialization. This plan should be based on existing vacancies (Identify a framework for school staffing based on the curriculum and student to teacher norms, in order to identify the needs for subjects by FTE), and projected teacher recruitment and compare the number of vacant positions to the number of available qualified and willing applicants (from the IARCSC database). The plan should identify districts with severe shortages of subject-qualified candidates (by gender) to be prioritized for teacher training services. 60. Improve data collection on teacher attrition (gender, age, location, qualification, reason for leaving), teacher vacancies (location, gender, subject), and applications, in order to utilize this data for teacher policy and management improvements. 61. Develop a comprehensive, up-to-date, accurate electronic database of all MoE managed teachers (and other employees). Integrate existing data (GDHR, EMIS, Payroll) into a single HR application that enables quick information extraction, administration of HR related management Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 22 files and macro (national) and provincial level reporting and analyses. The integrated system should include employees profiles, hours worked, leave days and salary payments, for permanent and temporary employees. Recommendations – Teacher Needs and Allocation The Ministry of Education should: SHORT TERM: (within the next 12 months) 62. Revise existing norms on teacher (and school staffing) needs per school. Provide clear and detailed instructions for schools, DEDs and PEDs to calculate the number of teachers needed, and identify the order in which type of teachers (subject specialization) should be prioritized, using a step-by-step model with progressively increasing student numbers. Criteria for creating classes should be clarified, as well as minimum and maximum benchmarks for class size. Norms for additional school staff (both academic and non-academic) should also be clarified in a similar step-by-step model. Furthermore, teacher (subject) qualification requirements per grade and school type should be specified. Revised norms should be clearly communicated to all levels of the public education system. 63. Apply a transparent and simplified formula for allocating teachers per province, using number of attending students per grade as the primary driver, and create an adjustment index for the percentage of rural schools (if needed). Develop instructions for PEDs and DEDs to prioritize intra-province distributions of new tashkeel positions, using relative teacher deficiencies as a driver. Publish the formula and methods used to allocate teachers across and within provinces and publish their outcomes, including PTRs, and where temporary teacher are considered. 64. Develop a policy and clear regulations for the contracting of haquzama teachers . Identify the purpose of this modality, and identify financial regulations and reporting requirements. Centralize planning and allocation of haquzama teachers, based on teacher needs assessments and sudden fluctuations in teacher needs. Utilize the flexibility of these contracts to address short term fluctuations, as well as work towards medium-term equity of the teacher distribution across provinces and schools. 65. Revise the existing policy toward very small classes. Increase the minimum benchmark, or allow for multi-grade classes in small rural schools. Furthermore, enforce regulation, so that very small classes are only created in rural areas to promote access to education, and not in semi-rural and urban areas. 66. Address PTR discrepancies between provinces and between schools, using the revised norms to identify excess and deficiencies. In the short term, relocate tashkeel teachers wherever possible, and use temporary teacher contracts to address severe deficiencies (i.e. focus on enormous classes). MEDIUM TERM: (within the next 36 months) 67. Progressively address PTR discrepancies between provinces and between schools. In the medium term, using attrition, recruitment and the allocation of new tashkeel position to address discrepancies. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 23 Recommendations – Teacher Recruitment The IARCSC should: SHORT TERM: (within the next 12 months) 68. Clearly describe the civil servant recruitment process, identify steps and regulations in order to improve transparency, accountability and increase awareness. This description should be published by the IARCSC and presented to its main stakeholder (the line ministries) as well as outside stakeholders (e.g. civil society). The IARCSC should clearly identify where, when, and for how long announcements, documents, results and complaints will be made available, and enable stakeholders to hold it accountable. 69. Develop standardized instructions for applicants to be shared during the announcement and application phases, which explain to applicants how processes work (application, examination, selection). The instructions should identify requirements, dates, time, and duration, as well as what mechanism exist for complaints and redress. 70. Thoroughly evaluate to ongoing recruitment round to better understand the validity and the fairness of the new exams, the impact of the chosen selection method (single non- transferable application), and its effectiveness of reaching target groups like women and rural applicants. Outcomes of the evaluation should be discussed with its main stakeholders (the line ministries) and ‘lessons learned’ to be implemented in the next round, should be identified. MEDIUM TERM: (within the next 36 months) 71. Develop transparent formulas/polices to integrate preferential treatment into the competitive recruitment process, to promote the recruitment and selection of target groups (e.g. women, rural applicants, applicants with certain subject qualifications), based on consultations with the line ministries. The Ministry of Education should: SHORT TERM: (within the next 12 months) 72. Clearly and transparently describe how the new teacher recruitment process works and communicate this to applicants and stakeholders. Prospective teacher applicants should be informed during the announcement and application phase of the process and regulations on applications (apply for one position only), biometric registration (the location and time period available), examination, and selection. 73. Identify target groups for preferential treatment in the competitive recruitment process, including women, rural teacher applicants, and applicants with scares subject specializations, and advise the IARCSC on how to advance their chances of being selected. 74. Use the IARCSC database on applicants, and exam results to develop a detailed and comprehensive overview of HR availability, and discrepancies in gender and (subject-) qualifications across provinces and districts. The overview should be used to inform teacher training needs, targeted recruitment policies and regional or gender based allowances. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 24 Recommendations – Teacher Remuneration and Costs The Ministry of Education should (jointly with the IARCSC and the Ministry of Finance): SHORT TERM: (within the next 12 months) 75. MoE should participate in the national dialogue on public sector remuneration led by the IARCSC and Ministry of Finance officials. This dialogue should focus on resolving the challenges of regional pay disparities, declining real salaries, and integration of incentives in the remuneration system. Participation in this process should seek to address these challenges within the medium-term fiscal envelope. The creation of a Civil Service Pay Commission, led by MoF and IARCSC, could facilitate policy dialogue on remuneration reforms. A teacher pay taskforce within the commission could look more in depth into specific challenges affecting teacher management. 76. Develop a more predictable education budget process (within the medium-term fiscal framework) to enable planning for the MoE, as well as PEDs, DEDs and schools. Reviewing expenditures relative to GDP would enable the MoE to more sustainably plan the wage bill. PEDs with indicative multiyear budget commitments will be empowered to develop multi-year plans to address shortages in pre-service trained applicants and improve the teacher GPI. 77. Develop a policy note that transparently identifies and illustrates the MoE’s fiscal requirements, and the effects of limited budget growth over the past years. The policy note should mobilize consensus across government and development partners for sustained growth of the education budget and advocate for a minimum percentage growths based on student enrollment, costs and inflation adjustments. MEDIUM TERM: (within the next 36 months) 78. Work progressively towards norm-based budgeting with per student allocations to provinces, districts and schools. This will require the delegation of more budget authorities to the provincial education directorates. Provinces with relatively low per student expenditures currently, will progressively receive more budget. PEDs need more flexibility to decide how excess funds can be spent, to stimulate progress in teacher shortages, quality, training and gender. A clear policy, regulations, and strong oversight are required to identify eligible expenditures for these provinces, to promote efficiency and prevent abuse. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 25 1 Introduction 79. This study aims to present a comprehensive analysis of teacher management issues in Afghanistan, in order to inform the Ministry of Education’s formulation of a new teacher management policy, and to build consensus amongst education development partners on priorities for reform and assistance in the short (within the next 12 months) and medium (within the next 36 to 48 months) terms. This report has a strong focus on quantitative data to analyze public management related issues including planning, human resources and financial management. It does not explore intrinsic teaching characteristics (teaching skills, methods, approaches). Rather it aims to inform the strengthening of teacher management to improve education outcomes. 80. This study has used a historical approach to teacher management issues, describing past developments to inform future priorities. A recurrent theme is the analysis of teacher management information from the past seven fiscal years. The report frequently presents historical data on planning, HR and finance in order to assess the effectiveness of policies and the degree to which strategic objectives are being met. Wherever possible, the analyses aim to forecast key indicators, such as teacher needs, training needs, and teacher costs, into the next seven years. In addition, models are presented to illustrate what developments are required over the next years, to achieve teacher management objectives (e.g. improving the teacher gender parity index). 81. The Ministry of Education has made tremendous progress in its capacity to manage the teacher workforce, despite the multitude of challenges it faced. Student enrollment in Afghanistan has increased exponentially, from 800,000 in 2002, to over 9 million today. In tandem, the Ministry of Education’s responsibilities, capacities and resources have grown. Each year, the Ministry needed to identify thousands of new teacher positions and recruit suitable candidates. Early on, the exponential growth combined with a lack of qualified teacher candidates, posed a tremendous challenge to the promotion of teacher quality and educations services. To address these issues, the MoE established teacher-training facilities and quickly expanded its geographic scope and training capacity. In 2016 alone, more than 23,000 people graduated from either pre-service or in-service teacher training. The MoE now manages almost a quarter of a million teachers, of which 33% are female, spread over more than 17,500 schools, including monthly salary payments, annual promotions and retirement. This is an achievement that deserves to be highlighted. 82. However, not all of Afghanistan’s provinces have witnessed improvements and growth evenly, and in part this reports describes how unequal progress has been. Some provinces have truly been deprived of most progress made elsewhere. These provinces typically have almost no subject-qualified teachers and almost no female teachers. The number of students per teacher is high, and expenditures per student are almost half that of the national average. While some of these inequalities have external causes (cultural norms, poverty, insecurity) this study shows that current teacher planning and management practices do little to address these inequalities, perpetuate them, or even amplify them. 83. The results of this study call for a more diversified teacher management approach that considers regional, gender and urban vs. rural discrepancies, in order to promote equity, efficiency and effectiveness. Standardizing teacher needs assessments, and a combination of class-based and top-down planning procedures will reduce the existing discrepancies in pupil- teacher ratios, whereas targeted teacher training plans and incentive schemes can address the Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 26 shortage of qualified and female teachers. Moving from an input –based to a norm-based budgeting process would result in surplus funding to the most deprived provinces, enabling them to address key constraints and catch up with progress made elsewhere. 84. Chapter 2 (Teacher Statistics) describes key teacher statistics, including their numbers, gender and qualifications. It also assesses teacher turnover, vacancies and applications, and shows how diverse these indicators are across provinces. The following three chapters look at the Ministry’s teacher management tools to address quality and equity. Chapter 3 (Teacher Needs and Allocation) describes how the MoE determines teacher needs, and how teachers are distributed across provinces and schools. Chapter 4 (Teacher Recruitment) describes how teachers were recruited in the past and assesses the recently introduced reforms in teacher recruitment. Chapter 5 (Teacher Remuneration) describes teacher remuneration, including base salaries, allowances, deductions and pensions. Last, chapter 6 (Teacher Costs) analyses per student expenditures, teacher costs, and forecasts the Ministry of Education’s wage bill until 2025. 85. Chapters 2 to 6 each end with a discussion and recommendations. As mentioned earlier, wherever possible these chapters aim to present developments in the past seven years, as well as forecasted or anticipated developments in the coming seven years. In addition, regional comparisons will be presented whenever this information is available and relevant. Methodology and Demarcation 86. This study combined quantitative data analyses on existing datasets with qualitative interviews. Between October 2017 and March 2018, 64 interviews were conducted with Ministry of Education officials and educations stakeholders. These included MoE officials from the planning and policy general directorate (GD P&P), the Education Management Information System (EMIS) unit, the payroll registration unit, the budget unit, the academic supervision directorate, the teacher education directorate (TED), the general directorate of human resources (GDHR), provincial education directorates (PEDs), school principals and teachers. Furthermore, interviews were conducted with officials from the Independent Administrative Reform and Civil Service Commission (IARCSC), the Ministry of Higher Education (MoHE), the Central Statistics Office (CSO), private schools, Non-Governmental Organizations (NGOs) and consulting firms involved in the education sector. 87. Datasets were retrieved from databases between October 2017 and March 2018. This study relied heavily on the payroll registration database, from which most teacher statistics were retrieved. EMIS data is also widely used, mostly in the context of analyzing student numbers in comparison to teacher or expenditure data. MoE expenditure data was retrieved from the Afghanistan Financial Management Information System (AFMIS). This study also used the IARCSC datasets on applicants, as well as tashkeel positions, and the GDHR datasets on attrition and allocation of new positions. MoHE and TED data on teacher training enrollment and graduation were retrieved, as well as a TED survey on teacher qualifications (sometimes referred to as TEMIS). The budget unit and academic supervision directorate provided additional datasets on haquzama (temporary contract) teachers. 88. Importantly, not all datasets collected for this study appeared consistent and complete. Datasets frequently appeared incomplete (e.g. GDHR attrition) inaccurate (EMIS student and teacher data) or conflicted with other datasets (haquzama data). These issues could have been caused by reporting or data entry error, or variations in reporting mechanisms. Data availability and the quality of databases is an important issue for teacher management itself, and discussed Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 27 in more detail in Box 2. Similarly, not all officials interviewed for this study were certain about official rules or processes, and occasionally they contradicted each other. The research team invested considerable time and effort to crosscheck statements and analyze the accuracy of large datasets, however it cannot fully guarantee data reliability. Wherever inconsistencies or conflicting data are relevant, they are discussed in the report (see for example Box 1, Box 2, Table 9 and Table 10). 89. Last, this study focused exclusive on teachers directly managed, administered and paid by the Ministry of Education, and does not cover teachers working in community-based education (CBE), private education, or NGO delivered education. These are systems regulated by the MoE but not implemented or directly managed. All these systems have an important symbiotic relation with MoE education services, especially CBE as it regulated by public policy, and functions as an extension of public education services in rural areas. However, implementation and management of CBE is fragmented, and a comprehensive analysis of CBE teacher management was beyond the scope of the study. Photo 2: Mixed boys and girls class (Kabul City, 2018) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 28 2 Teacher Statistics 90. This chapter discusses key teacher statistics. It aims to summarize information on teachers retrieved from large MoE datasets, to inform policy and decision-making7. The statistics in this chapter also provide context to the remaining chapters in this report. It discusses the number of employed teachers, teachers by gender, teacher qualifications, teacher attrition, and teacher vacancies and applications. Wherever possible, an overview of historical development on these indicators will be given, and regional differences will be described. This chapter ends with a set of recommendations. 2.1 Employed Teachers 91. There is no accurate, centralized and up-to-date registry of all teachers employed by the MoE. There are multiple databases on teachers, but not all are frequently updated, and none of them covers all types of teacher contracts. Therefore, this study was unable to determine the number of people employed as teachers by the Ministry of Education, or the number of teacher full time equivalents (FTEs), with full accuracy. 92. The most comprehensive and accurate database on employed teachers is the payroll registration database. This database is updated monthly based on the provincial education departments’ submission of their payroll registry, and lists all people receiving salaries from the MoE. The EMIS database, used for planning, reporting and analyses, uses the data from the payroll registration database as its primary source on teachers. However, the payroll registration unit does not keep track of vacant teacher positions, nor is it able to produce complete information on temporary teacher contracts. Other databases on teachers (including the ones maintained by the MoE GDHR, and the IARCSC) typically reflect the number of approved staff positions (i.e. the Tashkeel allocation) rather than the number of employed teachers8. 93. The Ministry of Education uses six different contract categories to employ people . The six different categories are described in Table 1. For more details on the different types of contracts, please see Appendix 1. Out of the six contract categories, five contract types are used to employ teachers. Teachers are employed through tashkeel teacher, haquzama inside tashkeel, and haquzama outside tashkeel contracts, but also both the ajeer (support staff) categories. Therefore, it is not clear how many teachers work for the MoE, as not all teachers are labeled as such. In addition, not all data regarding the different categories could be retrieved from central databases. For example, the payroll registration unit was unable to produce data on the haquzama inside tashkeel teachers. Last, some datasets on contract categories show conflicting numbers9 (typically haquzama data), and not all datasets are similar in scope and detail. While this could partly be attributed to different time frames, it is also likely caused by different reporting and payment mechanisms. 7 Data discussed in this chapter covers MoE administered teachers only, i.e. unless explicitly stated, it does not include community-based education (CBE), private education, tertiary education administered by the Ministry of Higher Education, and off-budget education service providers. 8 The ongoing recruitment of civil servants (early 2018) by the IARCSC created a much more detailed database that includes biometric data and detailed information on applicants. 9 The payroll registration database reported 13,690 haquzama outside tashkeel teachers, while the finance department reported 14,620. While the national difference is not significant, provincial numbers varied dramatically from 1 in Parwan (payroll registration database) to 1,528 in Parwan (budget unit). Importantly, individual PEDs reported significantly higher numbers of haquzama teachers than those registered in the payroll registration database. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 29 Table 1: Overview of different types of MoE employment contracts The bulk of all teachers employed by the MoE has this type of contract. It is a permanent civil servant contract specifically used for Tashkeel teachers. However, some staff working at larger high schools (e.g. I Teacher: principals, librarians) also have this type of contract. Although their number is relatively small, it thus means that not all tashkeel teacher contracts are actually used to employ teachers; This contract category is used for all MoE staff (except support staff) Administrative working at the central ministry in Kabul, the PEDs and DEDs. II Teachers are not categorized as administrative staff, but some Staff: administrative staff (e.g. academic supervisors) do function as teacher coaches and supervisors; This contract category is typically used for (lower graded) permanent civil servant support staff (cleaners, cooks, guards, III Ajeer Tashkeel: drivers, tea makers, etc.) in both the MoE offices as well as schools. However, many employees working as teachers are also categorized as Ajeer Tashkeel, usually when they do not meet the minimum education qualification of a teacher (grade 14); This contract category is used for the same type of employees as Ajeer ajeer tashkeel, however it is not a tashkeel (i.e. permanent civil IV servant) position, but a temporary contract instead. Similarly, some Bilmakta: teacher might be employed under this category. However, the overall number of people employed in this category is very low; Haquzama This is a contract category used to hire teachers (and other V Outside academic staff) temporarily. It is not a permanent civil servant Tashkeel: appointment, and used exclusively to hire people who are not also actively employed as a civil servant; This is a contract category used to hire teachers who are already Haquzama employed through a tashkeel teacher contract. It is a temporary VI Inside contract for a whole teacher FTE on top of the full time civil servant Tashkeel: contract. A haquzama inside tashkeel teacher thus has two full time teacher contracts. (Source: Interviews) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 30 94. There are approximately 218,261 people employed as teachers by the MoE10. This includes 203,641 tashkeel teachers11, and an additional 14,620 temporarily contracted teachers (haquzama outside tashkeel). The actual number of tashkeel could be 1-3% higher, given that some teachers work under an ajeer contract. Furthermore, an estimated 21,204 tashkeel teachers have a double full time contract (haquzama inside tashkeel) and the total number of teacher FTEs is estimated at 239,465, see Figure 1. The exact number of haquzama teachers could not be determined with full accuracy, see Box 1. Figure 1: Number of MoE employee contracts per category 2017 203.641 35.033 21.204 18.661 14.620 1.690 Tashkeel Outside Inside Haquzama Ajeer Tashkeel Ajeer Bilmakta Administrative Teachers Haquzama Teachers Staff Teachers 218,261 Teacher Employees Double Full-time contracts Mostly Support Staff 239,456 Teacher FTEs Possible some teachers incorrectly No Teachers labled (Source: Payroll Registration Database December 2017, MoE Budget Unit) 95. The number of tashkeel teachers has plateaued in the past five years (2013-2017), see Figure 2. While the number of teachers rose by almost 9% between 2011 and 2013, it thereafter remained stable (+/- 1%). This is due to an official policy directive preventing the hiring of additional tashkeel teachers in this period. The number of tashkeel teacher is expected to rise by 4% in 2018, due to the recruitment of an additional 8,000 tashkeel teachers. 10 These include all teachers employed in the MoE’s five programs: General Education, Islamic Education, Teacher Training, TVET and Literacy. 11 This number was retrieved from the December 2017 payroll registration database. In January 2018, the payroll registration unit categorized a significant number of tashkeel teachers as ajeer. The unit indicated that these employees were teachers, but working under an ajeer contract. Since it does not affect the number of employees working as teachers (but rather the number of contracts per category) the December 2017 database is used to facilitate multi-year comparison. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 31 Figure 2: Number of tashkeel teachers 2011-2017 201,726 203,491 203,535 200,387 203,641 191,754 185,255 1390 1391 1392 1393 1394 1395 1396 2011 2012 2013 2014 2015 2016 2017 (Source: Payroll Registration Database) Box 1: Haquzama contracts, the ‘hidden’ teacher workforce Haquzama contracts are a modality used to hire additional teachers through temporary contracts. Haquzama teachers are hired outside the tashkeel workforce (also known as: haquzama outside tashkeel) as well as inside the tashkeel workforce (also known as: haquzama inside tashkeel), thus providing an existing tashkeel teacher with an additional contract. Haquzama contracts are used for both teachers as well as academic managerial and support staff (e.g. principals and librarians). Teacher contracts are always full-time (24 hours per week) but contracts for academic support staff are often part-time (e.g. an additional 6 or 12 hours) and a substitute for additional work at schools with multiple shifts. Unlike tashkeel positions, the allocation across provinces, and the number of haquzama contracts, is not tightly regulated. Reporting on the number of haquzama contracts is also less systematic, less standardized and not comprehensively collated at the central MoE. The Ministry also does not publicly report on haquzama teachers, and none of its policy and strategic documents present figures on the haquzama teacher workforce. After an initial data request from the research team, the payroll registration unit provided data on tashkeel and ajeer bilmakta employees only. EMIS data on teachers is retrieved form the payroll database but never includes any indicators on haquzama teachers. Hitherto, haquzama teacher numbers have not been reported (periodically, transparently and publicly). Hence, many stakeholders are unaware of the existence or relative size of the haquzama teacher workforce. The research team eventually found five different sources on haquzama teachers, however all appeared incomplete or inaccurate: The payroll registration unit provided data on outside haquzama contracts (total 13,690) but was unable to provide haquzama inside tashkeel data; The academic supervision directorate provided data on haquzama contracts both inside and outside (total 31,760), but only for 23 provinces. Moreover, the dataset contained calculation errors and apparent data entry errors (e.g. more than 5,000 female haquzama teacher in Ghazni). Moreover the database included another 3,907 haquzama contracts for non-teacher academic staff; The budget unit provided a list of haquzama inside and outside contract numbers per province (total 35,824), but these were vastly different from the payroll numbers; Two PED officials interviewed for this study reported significantly higher numbers of haquzama contracts than those found in any of the datasets; The AFMIS database was consulted to estimate haquzama contracts based on expenditure data per province. Haquzama salaries are paid out of the ‘overtime’ budget line. A key problem is that ‘regular’ overtime payments (of DED and PED staff) are also paid out of this budget line (and it not possible to separate the data). While this should be a relatively small portion (given the size of this group of employees) it prevents accurate estimates. Table 10 presents haquzama numbers (inside and outside) from the three different datasets as well as estimate based on AFMIS data. Ultimately, accurate nation-wide or provincial figures could not be determined. Although some trends are discernable, the datasets differ too much to approach a reliable estimate. However, it is important to note that the AFMIS estimates, the PED interviews and the academic supervision dataset suggest that the actual number of haquzama teacher contracts is higher than the 35,824 reported by the budget unit, and reproduced in this report. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 32 96. It is not possible to determine the exact number of haquzama teacher FTEs. There is also no multi-year data available on haquzama teachers (both inside and outside tashkeel). However, the historical trends of haquzama teachers can be estimated using data from AFMIS. The estimates show a doubling of haquzama contracts between 2011 and 2013 and thereafter stabilize around 35,00012 (see Figure 3). Note that the actual number of haquzama teachers could be higher, as discussed in Chapter 0. Figure 3: Conservative estimate of number of haquzama contracts (inside and outside tashkeel) 2011-2017. 2010- 2016 estimates are based on AFMIS analyses. 45.779 44.558 41.676 38.922 35.824 26.084 20.301 1390 1391 1392 1393 1394 1395 1396 2011 2012 2013 2014 2015 2016 2017 (Source: AFMIS) 97. Between 2011 and 2017, the number of enrolled students increased faster than the number of tashkeel teachers and haquzama contracts. As a result, the pupil-teacher ratio (PTR), dividing the number of enrolled students by the number of tashkeel teachers, rose from 41 in 2011 to 46 in 2017.13 However, the development of PTRs is altered somewhat if the estimates of haquzama teachers are taken into account (from 37 in 2011 to 39 in 2017). Furthermore, Afghanistan’s MoE uses a fairly unique definition of ‘enrolled’ students (students remain registered as ‘enrolled’ even when absent for almost three years). Using ‘attending’ students as an indicator for calculating PTRs would better reflect actual conditions14. The net PTR (attending students per teacher FTE) rose from 28 in 2011, to 30 in 2017 (see Figure 4). For a more detailed discussion on regional PTRs and teacher needs, see chapter 3. 98. The number of tashkeel teachers varies considerably across provinces. There are more than 27,000 tashkeel teachers in Kabul City15, and Badakhshan, Balkh, Nangarhar and Herat provinces each have between 10,000 and 15,000 tashkeel teachers. Combined, these five education directorates employ 40% of the tashkeel teacher workforce. Uruzgan, Nuristan, 12 Haquzama inside and outside tashkeel teacher are paid out of the ‘overtime’ expenditure code in AFMIS. This code also includes overtime payment of MoE administrative staff. This method used annual expenditures on ‘overtime’ relative to the size of tashkeel teacher salary expenditures, using 2017 (with known numbers of haquzama) as an index to extrapolate. Given that the size of overtime expenditures of MoE administrative staff are unknown, actual numbers might deviate somewhat from these estimates. 13 This calculation uses all enrolled students (all 5 MoE programs) and all tashkeel teachers (all 5 MoE programs). 14 Data on student attendance could not be retrieved for all years (2011-2017). The calculation presented uses a fixed attendance rate of 76%, as was retrieved form 2017 EMIS data, and extrapolated to previous years. 15 Kabul City is administered by a provincial education directorate (PED) separate from Kabul Province. As a result, there are 34 provinces in Afghanistan, but the MoE has 35 PEDs. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 33 Nimroz, Panjshir, and Zabul provinces each employ between 1,000 and 1,600 tashkeel teachers, and combined these directorates employ only 3.5% of tashkeel teachers. For an analysis on how these tashkeel teacher numbers compare to enrolled student numbers please see page 62. Figure 4: PTR development 2011-2017 (students of all five programs) using a variation of indicators 46 46 45 44 42 42 41 39 39 39 37 37 36 35 35 35 34 33 32 32 31 29 30 30 28 28 27 27 1390 1391 1392 1393 1394 1395 1396 2011 2012 2013 2014 2015 2016 2017 enrolled students to tashkeel teacher ratio attending students to tashkeel teacher ratio enrolled students to total teacher FTE (est) ratio attending students to total teacher FTE (est) ratio (Source: EMIS, PAYROLL, and Budget Units) 99. Provincial data on haquzama contracts is less reliable (different MoE units reported different numbers per province, for 2017). According to the payroll registration database, Kandahar (1,439), Kabul Province (1,467), Khost (1,777), Kabul City (2,095) and Herat (2,207) employ two-thirds of all haquzama outside tashkeel teachers. Whereas many provinces (Daikundi, Nimroz, Kunduz, Paktia, Parwan, Samangan, Farah, Uruzgan) employ less than three haquzama outside tashkeel teachers. However, the MoE budget unit reported 1,528 haquzama outside tashkeel teachers in Parwan (instead of zero in the payroll database), and some PEDs interviewed for this study reported haquzama numbers considerably higher than those registered in the central ministry databases. The discrepancies between the numbers of haquzama teachers could be the result of different provincial reporting mechanisms of these types of contracts16. For a more detailed discussion on the provincial distribution of teachers, see Chapter 0. For an overview of data collected on haquzama teachers per province, see Table 10. 16 One payroll unit official indicated that not all provinces record haquzama teachers in the payroll registration database. In addition, PED officials indicated they reported haquzama data to the general education (academic supervision) and planning general directorates. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 34 2.2 Teachers by Gender 100. In 2017, the MoE employed 68,113 female tashkeel teachers, and 135,528 male tashkeel teachers17. Thus, 33% of tashkeel teachers are female and 67% are male. Converted, the gender parity index (GPI: female divided by male indicator) for tashkeel teachers is 0.50 (see Figure 5). The GPI for haquzama outside tashkeel teachers is somewhat higher, at 0.69 (gender segregated data for haquzama inside tashkeel teachers could not be retrieved for this study). The relatively high female participation in the teacher workforce18 is compatible with global trends, where, in culturally conservative societies, women’s employment in the public education sector is more accepted than other sectors. The GPI for other MoE contracts (administrative staff, and both ajeer categories) is significantly lower, at 0.11, reflecting the limited number of women hired in managerial and support staff functions. Figure 5: MoE employees by contract category and gender 2017 90% 67% 59% 41% 33% 10% Male Female Male Female Male Female Tashkele Teacher GPI Haquzama Outside Tashkeel Other MoE Contracts .50 GPI .69 GPI .11 (Source: Payroll Registration Database and Budget Unit) 101. In the past seven years, the GPI for tashkeel teachers has improved somewhat (see Figure 6). In 2011 there were only 44 female tashkeel teachers per 100 male teachers (GPI 0.44), in 2017 this number had risen to 50. However, at the current pace the MoE is not reaching its targets of promoting female teacher recruitment. In the NESP III, the MoE has expressed its commitment to promoting female teacher recruitment as a means to promoting girls’ enrolment and access to education. The NESP III indicates that the Ministry aims to increase the teacher gender parity index to 0.69 by 2021.19 In other words, the MoE aims to increase the number of female teachers to 69 per 100 male teachers. However, in the past six years (2011- 2017), this number increased from 44 to 50 female teachers per 100 male teachers. At the current pace, this target will not be achieved. 17 Payroll Registration Database, December 2017. 18 The GPI for Afghanistan’s entire civil service workforce is 0.23 (World Bank staff calculations), significantly lower. 19 Specifically, the NESP III indicates the MoE’s target of increasing the percentage of female tea chers to 40,7%, the equivalent of a 0.69 GPI. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 35 Figure 6: GPI development tashkeel teachers 2011-2017 0.50 0.50 0.48 0.46 0.47 0.44 0.44 1390 1391 1392 1393 1394 1395 1396 2011 2012 2013 2014 2015 2016 2017 (Source: Payroll Registration Database) 102. It is important to stress here that the MoE has limited opportunities for improving the GPI, given the permanent nature of civil servant contracts, limited attrition in the teacher workforce (see 2.4 Teacher Attrition), and limited growth of the teacher workforce in recent years. A significant increase in the tashkeel teacher GPI, as envisioned in the NESP III, would require drastic policy measures. Assuming the MoE will continue to increase the number of tashkeel teachers by around 8,000 each year, at least 85% of new teacher recruits should be female in order to achieve this target, see Figure 7. Figure 7: Forecasting female teacher needs to achieve NESP III targets until 2021 135.528 136.791 138.055 139.318 140.581 96.487 89.393 82.300 75.206 68.113 0.50 0.55 0.60 0.64 0.69 1396 1397 1398 1399 1400 2017 2018 2019 2020 2021 male female GPI (Source: Staff Calculations) 103. Moreover, significant gender variations exist across provinces, and in around half of all provinces the tashkeel teacher GPI is still below 0.25, see Map 1. More than half of all female teachers in Afghanistan work in Kabul city, Balkh or Herat province, and the national average GPI hides extreme differences in the teacher gender gap between provinces. In the 17 provinces with the lowest GPI, there are on average only 12 female teachers per 100 male Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 36 teachers, and on the extreme end (Paktika, Khost, Kunar) the number is between 2 and 6. In the seventeen provinces with the highest GPI, the average number is as high as 62, and on the extreme end (Balkh, Nimroz, Herat) there are more female teachers than male teachers, see Table 2. Map 1: Afghanistan, number of female tashkeel teachers per 100 male tashkeel teachers, 2017 78 136 44 51 50 61 43 36 62 29 14 25 33 18 20 6 8 42 6 13 109 16 12 64 7 36 4 9 2 57 258 KABUL CITY 9 50 NATIONAL AVERAGE 22 23 135 (Source: payroll registration database) 104. Importantly, in the past seven years, those provinces with the least female teachers have made almost no progress in recruiting more female teachers. In these provinces, improvements in the GPI are marginal (+ 0.02), whereas in the provinces with the highest GPI progress is more significant (+ 0.11). Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 37 Table 2: Tashkeel Teachers by gender 1390 and 1396 Tashkeel teachers by gender 1390 1396 1390-1396 Province Male Female GPI Male Female GPI +/- Kabul City 6,578 17,716 2.69 7,614 19,681 2.58 - 0.11 17 provinces with highest GPI 65,448 33,340 0.51 66,370 41,025 0.62 + 0.11 Balkh 5,152 5,654 1.10 5,134 6,958 1.36 + 0.26 Higher end GPI Nimroz 656 718 1.09 610 826 1.35 + 0.26 Herat 7,926 6,922 0.87 7,597 8,248 1.09 + 0.21 Provincial Average (ex Kabul City) 122,021 38,940 0.32 127,914 48,432 0.38 + 0.06 17 provinces with lowest GPI 56,573 5,600 0.10 61,544 7,407 0.12 + 0.02 Paktika 3,451 26 0.01 3,337 58 0.02 + 0.01 Lower end GPI Khost 3,986 126 0.03 4,253 188 0.04 + 0.01 Kunar 3,767 172 0.05 3,884 252 0.06 + 0.02 (Source: Payroll Registration Database) 105. Some provinces, in particular in the North and West of Afghanistan20, are significantly increasing the number of female teachers through targeted recruitment. These provinces hire at least twice as many female teachers as male teachers, quickly working towards a teacher gender balance. In some provinces, the PEDs (appear) to have hired only female teachers21. However, in the 17 provinces with the lowest GPI, the PEDs are still hiring far less women than men. On average only 21 female teachers were hired for every 100 male teachers. 20 For example Nimroz, Herat, Daikundi, Bamiyan, Faryab, Balkh, Takhar, Farah, Baghlan. 21 Note that this analysis is based on estimates, and not actual data of new recruits. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 38 Table 3: Tashkeel Teacher Recruits by Gender 1390 and 1396 Tashkeel change Attrition NEW Recruitments22 Recruits (Tashkeel) by gender 1390-96 (EMIS) 1389-95 (GDHR) 1390-96 (EMIS+GDHR) Province Male Female Male Female Male Female GPI Kabul City 1,036 1,965 2,330 1,841 3,366 3,806 1.13 17 provinces with highest GPI 922 7,685 3,720 479 4,642 8,164 1.76 Balkh -18 1,304 469 144 451 1,448 3.21 Higher end GPI Nimroz -46 108 60 1 14 109 7.79 Herat -329 1,326 558 88 229 1,414 6.17 Provincial Average (ex Kabul City) 5,893 9,492 7,847 577 13,740 10,069 0.73 17 provinces with lowest GPI 4,971 1,807 4,127 98 9,098 1,905 0.21 Paktika -114 32 20 0 -94 32 - Lower end GPI Khost 267 62 201 1 468 63 0.13 Kunar 117 80 209 1 326 81 0.25 (Source: Payroll Registration Database and GDHR) 106. While it is clear that in at least half of Afghanistan’s provinces, the female teacher recruitment targets are not being met, it is unclear why the PEDs with the least number of female teachers are not hiring more women. Accurate data on recruits as well as applicants would allow to better assess, if this issue is caused by a lack of qualified female teacher applicants in these provinces, if the PEDs are not effectively prioritizing female recruitment, or both. 107. If the MoE wants to address the significant teacher (and student) gender gap in these provinces, it needs to clearly develop a recruitment policy that specifies how gender is prioritized in the teacher recruitment process, in order to increase the rate of female teacher recruitment and achieve its targets. 2.3 Teacher Qualifications 108. The minimum education qualification of a tashkeel teacher (teaching grades 1 to 12 in primary, lower and higher secondary) is ‘Grade 14’, the equivalent of high school plus two additional years (usually completed at a teacher training college). Teachers with a ‘grade 14’, ‘undergraduate’ (grade 16 university bachelor’s), Master’s, or PHD, are considered qualified teachers. Whereas teachers with only a primary, lower secondary or high school diploma are considered under qualified. Historically most teachers employed by the MoE were under 22 The Ministry of Education does not centrally collect information on ‘ new recruits’. Calculating new recruits based on tashkeel total number changes (EMIS) and attrition (GDHR), assuming those databases are accurate, remains an estimate containing error, as it does not reflect teacher inter province relocation data. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 39 qualified. MoE recruitment practices allowed for hiring under qualified tashkeel teachers in exceptional circumstances (i.e. in those regions where there are not enough qualified teacher applicants)23. 109. According to EMIS, the number of qualified tashkeel teachers almost doubled in recent years, from 61,091 in 2011 to 114,762 in 2016.24 And the percentage of tashkeel teachers with a grade 14 (or higher) qualification rose from 33% (of total tashkeel teacher) in 2011, to 57% in 2016 (see Figure 8). In 2016, for the first time, there were more qualified tashkeel teachers than under qualified teacher employed by the MoE. The significant rise in the numbers of ‘grade 14’ educated teachers (an increase of 53,000 between 2011 and 2016) cannot be attributed to recruitment and attrition alone. A likely explanation is that the considerable number of teachers that received in-service teacher training (83,963 in-service teachers between 2010 and 2016) was afterwards registered as ‘grade 14’ in the payroll registration database. Figure 8: Development of tashkeel teacher qualification 2011-2016 33% 38% 43% 45% 67% 50% 62% 57% 57% 55% 50% 43% 1390 1391 1392 1393 1394 1395 2011 2012 2013 2014 2015 2016 n= 185.255 n= 193.619 n= 201.982 n= 206.527 n= 203.681 n= 200.311 Below Grade 14 Grade 14 or above (Source: EMIS database) 110. The percentage of female teachers that meets the minimum ‘grade 14’ qualification level, is higher than for male teachers. Two-thirds of female teachers (66%) were ‘grade 14’ educated/trained in 2016, up from 45% in 2011. Whereas a little over half of the male teachers (53%) had this qualification in 2016, up from 28% in 2011 (see Figure 9). A potential explanation for the relatively high percentage of qualified female teachers is their concentration in large urban centers (Kabul, Herat, and Mazar-e-Shariff) where education levels are assumed to be higher, and in provinces with relatively high percentages of qualified tashkeel teachers. 23 Moreover, the lack of merit-based recruitment, and potentially abuse and favoritism during the recruitment process (see Chapter 4) might have led to higher numbers of under qualified tashkeel teachers. 24 The EMIS database on teachers retrieves information from the Payroll Registration Database. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 40 Figure 9: Development of tashkeel teacher qualification levels male/female 2011-2016 (No data for 2012) 100% 80% 60% 40% 20% 0% 1390 1391 1392 1393 1394 1395 2011 2012 2013 2014 2015 2016 Male Tahkeel Teachers primary secondary 12th grade 14th grade 100% 80% 60% 40% 20% 0% 1390 1391 1392 1393 1394 1395 2011 2012 2013 2014 2015 2016 Female Tashkeel Teachers (Source: EMIS) 111. However, the national averages hide dramatic differences in the qualifications of teachers across different provinces. In some provinces, including Paktika, Daikundi, Zabul, Kandahar, Uruzgan, and Badghis, less than 20% of teachers is ‘grade 14’ educated. Whereas in Kabul City, Balkh, Jawzjan, Faryab and Kapisa, more than 75% of teachers are ‘grade 14’ educated (in 2016). Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 41 Table 4: Qualified tashkeel teachers as percentage of total tashkeel teachers 2011 2016 +/- 90-95 Province Male Fem Total Male Fem Total Male Fem Kabul City 20% 51% 70% 24% 61% 85% + 5% + 10% 17 provinces with highest % 26% 10% 36% 48% 19% 66% + 22% + 9% Kapisa 47% 7% 54% 66% 13% 79% + 19% + 6% Higher end Jawzjan 23% 29% 51% 40% 36% 76% + 17% + 7% Qualification Laghman 43% 2% 44% 62% 5% 67% + 19% + 3% Provincial Average (ex Kabul City) 19% 8% 27% 37% 16% 53% + 18% + 8% 17 provinces with lowest % 11% 6% 16% 24% 12% 35% + 13% + 6% Daikundi 4% 0% 4% 9% 3% 12% + 5% + 3% Lower end Paktika 4% 0% 4% 10% 0% 10% + 6% + 0% Qualifications Ghor 5% 0% 5% 28% 3% 32% + 23% + 3% (Source: EMIS) 112. Similarly, in Faryab, Takhar, Bamiyan, and Wardak, the percentage of ‘grade 14’ educated teachers rose with more than 40% (suggesting considerable investments in teacher in-service training there). Whereas in Zabul, Nuristan, Paktika, Kandahar, and Badghis less than 10% increase in the ‘grade 14’ category was achieved. This suggests that the combined impact of new recruits and in-service training is resulting in an increase of the qualification gap between provinces. While nation-wide figures on ‘grade 14’ teachers have improved (+23%), the variation between provinces in the qualification of their teachers has become larger25 between 2011 and 2016, see Table 4. 2.3.1 Qualifications and Gender 113. The teacher qualification gap compounds the teacher gender gap (r=.47). The provinces with the highest percentage of female teachers, also tend to have the highest percentage of ‘grade 14’ female teachers. Kabul City and Balkh Province have GPIs of 2.58 and 1.36 respectively (i.e. both have more female than male teachers) and in both cases 84% of all female teachers is ‘grade 14’ (or above) educated. Conversely, Paktika has 3,648 teachers of which only 58 female teachers, but none of them has a ‘grade 14’ education. Wardak has 4,553 teachers, of which only 266 female teachers, but only 36 with a ‘grade 14’ education. 114. In fact, in Wardak, Paktya, Zabul, Uruzgan, Paktika and Khost, less than 1% of all tashkeel teachers is a ‘grade 14’ educated female teachers. And these provinces barely made any improvements on mobilizing ‘grade 14’ qualified female teachers between 2011 and 2016. Whereas in Kabul City 60% of all teachers is ‘grade 14’ educated and female, and in Balkh, Jawzjan and Faryab this is 47%, 36%, and 32% respectively. All these provinces improved this indicator significantly during the 2011-2016 period, see Map 2. This appears to suggest that the 25 Standard deviation (ST.DEV) increased from 15% in 2011 to 22% in 2016. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 42 impact of in-service teacher training (and potentially the resources therefore) are not evenly distributed across the country, and that the MoE needs to develop a more targeted approach to teacher training and female teacher recruitment to reduce this gap. Map 2: Percentage of qualified teachers 2016 (in color) and percentage increase 2011-2016 (in digits) 7% 19% 4% 10% 17% 18% 6% 5% 8% 8% 1% 4% 12% 5% 6% 1% < 20% 3% 9% 21%-33% 1% 3% 34%-46% 12% 4% 3% 47%-60% 3% 0% 61%-65% 8% 0% 66%-74% 0% > 75% 0% 8% 0% 10% KABUL CITY 8% NATIONAL AVERAGE 2% 2% 15% (Source: EMIS) 2.3.2 Subject Qualification 115. Provincial teacher qualification gaps are further amplified by discrepancies in subject qualifications. In addition to regional differences in the percentage of qualified and under qualified teachers (i.e. the teacher’s level of education enjoyed), there all also difference in the degree to which teachers are subject qualified (i.e. if the teacher studied the subjects that he or she teaches). A 2016 survey of the TED suggests that the existing discrepancies across provinces are amplified if subject qualifications are considered26. Provinces with few qualified teachers (Paktika, Daikundi, Zabul, Kandahar, Uruzgan and Badghis) have even fewer subject-qualified teachers that studied the subject they teach. 116. The study’s data shows there are considerable nation-wide differences between subjects. 67% of Islamiat teachers are subject qualified and 59% of science teachers are subject qualified, compared to 41% of English teachers and 40% of Pashto teachers. Within provinces, differences in subject qualifications are more pronounced. For example in Farah province, 99% of its Islamiat teachers are subject qualified, but only 31% of math teachers. In Ghor province, 87% of Islamiat teachers are subject qualified, but only 10% of Math teachers, 15% of Science teachers, 26Teacher’s Mapping & Training Projection Survey Report, August 2016, Teacher Education Directorate, Islamic Republic of Afghanistan. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 43 7% of social science teachers, and there are no subject qualified English teachers at all. The data shows that in provinces with very few qualified teachers, it is typically the Islamiat teachers that are subject qualified and for other subjects there are almost no subject qualified teachers. 2.3.3 Actual Teacher Abilities 117. Importantly, the average abilities of qualified teachers might differ across provinces, as there are no standardized national tests, for pre- and in-service trained teachers. Teacher training colleges (TTCs) and university education faculties do not apply the same standardized tests to all their facilities across provinces (even though the curriculum is often standardized). While the quality of training at these institutions is likely to vary considerably, a similar variation will exist in the abilities of ‘qualified teachers’. As a result, analyzing payroll registration data on the number of teachers with adequate education certificates, only partially reflects the discrepancies between teacher abilities across the country. 118. There is no nation-wide standardized testing of teacher abilities within the pool of MoE employed teachers. Although the teacher education directorate (TED) in Kabul indicated it uses tests during in-service training, there is no data or summary on the outcomes of these tests. There is also no comparative material on the graduates of pre- and in-service training, and the regular civil servant performance appraisal (regulated by the IARCSC but often performed by school principals) is widely perceived to be a ceremonial exercise, rather than an effective assessment. According to the GDHR, only 70 teachers (0.03%) did not reach the minimum score of the appraisal in 2017, and were thus recommended for training. As such, there is no comparable information on the actual abilities of teachers across regions, and across programs and school types (e.g. primary, lower and higher secondary). 119. A recent assessment of education service delivery in Afghanistan tested 1,598 teachers spread over 200 schools, on their language and numerical abilities27. The results of this study suggest that a significant portion of the teachers is severely under qualified. For example, teacher tested in the study lacked the reading comprehension skills that are expected of 4 th grade students, and teachers scored on average lower than the average score of 4th grade students internationally. Furthermore, half of the (primary school) teachers tested, was unable to divide double digits by single digits or calculate the perimeter of a rectangle. While these findings did not significantly deviate from a comparable pool of studies (in Sub-Sahara Africa) it does indicate the significant discrepancy of teacher abilities within the MoE employed teacher pool. 2.4 Teacher Attrition 120. Official data on teacher attrition appears to be incomplete or inaccurate. The GDHR keeps track of tashkeel teacher attrition across the country, based on information submitted by the PEDs’ HR departments. However, close inspection of the national MoE dataset on attrition (2010-2016)28 suggests that it is incomplete and contains errors. 27 The Learning Crisis in Afghanistan, Results of the SABER Service Delivery Survey 2017 (PRELIMINARY DRAFT). P.56- 63. The World Bank, 2018. 28 The GDHR shared three different files regarding attrition: (i) PDF and (ii) excel file providing a national summary of attrition data, and (iii) excel file with detailed information (per province per year and disaggregated by gender, reason for leaving, etc.). Each file presents different numbers on attrition, and the detailed dataset (iii) contains multiple internal errors (mutually exclusive categories adding up to more than the total). Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 44 121. First, some provinces reported extremely low attrition rates (e.g. Sar-e-Pul reported 0.36% attrition over a period of seven years) or no attrition at all (e.g. in Daikundi not a single person retired over a period of seven years). Second, the dataset contains internal errors (different categories do not add up to the total sum). Third, the dataset identifies four categories under ‘reason of retirement’ including (i) high age; (ii) long serving; (iii) ‘death’; and (iv) ‘own will’; suggesting there is no category for involuntary termination of contract, nor registration of inter-provincial relocation of teachers. And last, the average year-on-year attrition rate is below 1%, which (while technically possible) appears to be very low. Combined, this suggests that the GDHR dataset on attrition reflects only a partial picture of actual teacher attrition numbers across the county. While highlighting these issues, the following can be deducted from the GDHR attrition database. 122. 10,893 Tashkeel teachers ‘retired’ in the period between 2010 and 2016. On average 1,570 per year. This ‘retired teacher’ category includes multiple reasons for terminating active duty, including (i) ‘old age’; (ii) ‘long serving’; (iii) ‘death’; and (iv) ‘own will’. These numbers only reflect tashkeel teachers, and not teachers in the tashkeel ajeer category, or teachers with a temporary contract (haquzama and ajeer bilmakta). Figure 10: Tashkeel teacher attrition 2010-2016, in absolute numbers and as a percentage of the total 2,500 1.60% 1.40% 2,000 1,907 1,715 1,728 1.20% 1,484 1,480 1.00% 1,500 0.97% 0.95% 1,318 0.85% 1,261 0.80% 0.80% 0.73% 0.69% 1,000 0.62% 0.60% 0.40% 500 0.20% 0 0.00% 1389 1390 1391 1392 1393 1394 1395 2010 2011 2012 2013 2014 2015 2016 (Source: GDHR) 123. In the same period (2010-2016), the tashkeel teacher attrition rate fluctuated between 0.62% and 0.97%, see Figure 10. That is, each year, out of a thousand tashkeel teachers, between six and ten teachers retired from active duty. This appears to be a very low rate. If teachers were evenly distributed across age categories, the percentage of ‘old age’ and ‘long serving’ retirees alone would be around 2.5%. A potential explanation for the low attrition rates could be the age composition of the teacher pool, with many young teachers hired in the past Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 45 17 years that have yet to reach the retirement age (different MoE units were unable to produce data on the age composition of the tashkeel teacher pool). 124. The attrition rates for female teachers (between 0.42% and 0.72%) are somewhat lower than for male teachers (between 0.77% and 1.11%)29. Within the tashkeel teacher pool women represent 32% of employees (average 2010-2016) but only 22% of retirees. A possible explanation for lower female attrition rates could be that female teachers are on average younger (and recruited more recently) than their male colleagues30. Figure 11: Reason for retirement and education level of retiree 2010-2016 59% 39% 25% 21% 18% 15% 17% 7% High age Long serving Death Own will Below 12 Grade 14 Grade Grade 16 Grade 12 and above Reason for 'retirement' Education Level of Retiree (Source: GDHR) 125. Out of the entire ‘retiree’ category in the GDHR database (2010-2016), 60% of MoE employees reached ‘retirement’ (age 65 or served as a tashkeel teacher for 40 years). 25% of employees terminated their employment voluntarily before reaching ‘retirement’. And 15% passed away before reaching retirement. Furthermore, only 25% of retirees was ‘qualified’ (grade 14 or above), whereas in the same period on average 44% of teachers was qualified, suggesting a higher attrition rate amongst lower educated tashkeel employees. 31 Last, information on provincial attrition rates appears unreliable and will not be presented here. 2.5 Teacher Vacancies and Applications 126. There appears to be no centralized MoE database on teacher vacancies or applications . This study was not able to find a list of teacher vacancies or applications across provinces. However cross checking lists of the IARCSC on tashkeel (the approved civil servant staff structure) with the payroll registration database provides an initial indication of the amount of MoE and teacher vacancies. However, a primary risk of this method is that the analyses reflect database discrepancies rather than actual vacancies. 127. According to the IARCSC, there are 271,899 positions in the MoE approved tashkeel (2017). However, there are only 257,335 employees registered in the payroll database as 29 This is actually an estimate. The dataset does not specify gender for teachers retiring, but only gender of total attrition (including all MoE staff). However, given that the GPI for MoE administrative staff and Ajeer is much lower than for teachers, it is likely that almost all female MoE retirees are female teachers. 30 As the gender of retirees is only specified for the whole group, and not sub categories such as ‘reason for leaving’, it is not possible to analyze if female teachers are mostly retiring due to ‘old age’ or due to ‘own will’. 31 This is based on an analyses of all MoE retirees, rather than only teachers, as the database does not specify data fro teachers separately. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 46 tashkeel (this includes tashkeel teachers, administrative staff, and ajeer tashkeel, but excludes the 35,824 haquzama contracts, and the 1,690 ajeer bilmakta). This suggests that there were 14,564 vacancies within the MoE (in 2017), a little over 5% of the approved staff structure. Assuming vacancies are equally distributed across tashkeel contract types, there were an estimated 11,525 teacher vacancies across the country in 2017, or on average 329 tashkeel teacher vacancies per provincial education directorate32. 128. Based on a comparison of the IARCSC 2014 provincial tashkeel, and 2014 payroll registration data, with a few exceptions vacancies appear to be evenly distributed across most provinces (between 1% and 10% vacant positions). Four provinces have an exceptionally high percentage of vacancies, including Zabul (51%), Daikundi (49%), Panjshir (24%) and Badghis (19%). Whereas some of the PEDs with the most tashkeel teachers (Kabul City, Herat, Nangarhar) and Kandahar Province employ a few hundred more teachers than approved in their staff structure. However, it needs to be stressed that there is a risk that these numbers are inflated by database discrepancies. 129. At the time of writing (February/March 2018) the IARCSC was in the process of hiring an additional 17,700 civil servants, of which 8,000 teachers, through a new recruitment process (see Chapter 4). This new round of recruitment established for the first time a national centralized database on teacher (and civil servant) applicants. In late 2017, the IARCSC received over 250,000 applications, of which 225,322 were eventually shortlisted (after an initial screening for eligibility). The shortlist included 118,544 applicants for 8,000 teacher vacancies nation-wide. Thus, on average almost 15 people applied for every teacher vacancy published by the IARCSC. 130. The number of IARCSC shortlisted teacher applicants varies considerably across provinces. In Balkh, Nangarhar and Herat around 10,000 applicants were shortlisted, and in Kabul (City and Province) more than 22,000 applicants were shortlisted. However, in some provinces the number of shortlisted applicants for teacher position was significantly lower, including Zabul (94), Uruzgan (111), Nuristan (192), and Nimroz (283). The number of teacher applications received by the IARCSC correlates significantly with the number of TTC pre-service graduates, as will be discussed in Chapter Error! Reference source not found. (Teacher Training). 131. In some provinces, on average more than forty applicants were shortlisted for each available teacher position in that province, suggesting significant interest in the teacher profession there, as well as an adequate supply of eligible candidates33. The average number of applicants per vacancy was highest in Faryab (50.9), Baghlan (43.5), Parwan (40.3), and Kapisa (39.8). Conversely, in some provinces, on average less than two applicants were shortlisted for each available teacher position in that province, suggesting a significant shortage of willing and eligible candidates for teacher positions. Average shortlisted applicants were lowest in Uruzgan (1.1), Paktika (1.3), Zabul (1.6), Nuristan (1.9) and Daikundi (2.1) see Map 3. 32 This calculation predates the new round of teacher recruitment for an additional 8,000 vacant positions. 33 Unfortunately the IARCSC was unable to share exam results. This would provide a more accurate indication of the number of qualified and willing teacher applicants. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 47 Map 3: Shortlisted teacher applicants per vacancy by province 2018 28.3 29.2 27.2 24.5 16.0 50.9 11.9 43.5 9.4 6.1 1.9 3.0 13.0 40.3 39.8 12.1 17.5 24.1 19.3 13.5 22.3 5.3 4.8 2.1 6.3 7.1 5.0 1.1 1.3 18.6 1.6 14.0 3.9 4.6 (Source: IARCSC 2018) 132. Importantly, only a section of shortlisted applicants came to pick up their exam ID card and submit their biometric data for registration. In four southern provinces (Kandahar, Zabul, Uruzgan, and Nimroz) only 60% of shortlisted candidates were subsequently recorded in the biometric database, and in Uruzgan Province only 49% was registered34. As a result, there are less registered teacher applicants in Uruzgan than teacher vacancies . And this analysis only considers the total number of applicants and vacancies, and not their location (within the province) or the type of vacancy (e.g. chemistry teacher), which would stress an even grimmer picture. The data clearly shows that finding teachers (for specific positions in specific villages) in Uruzgan is severally challenged, by the limited number of qualified and willing candidates. 133. Regrettably, the IARCSC was unable (at the time of writing) to share gender-disaggregated data on the applicants of the new positions or the vacancies themselves. This would allow identifying regions where female applicant shortages are, and could be used in the design of targeted policies to mobilize women in the teacher workforce. Similarly, this study was unable to retrieve exam outcomes. 34 IARCSC press release 17th February 2018, retrieved from the IARCSC website (February 27th 2018). Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 48 Box 2: Teacher Databases, Data Gaps, and Data Integration Teacher Databases Payroll Registration Database – widely considered the most comprehensive and accurate database on teachers currently available. It’s the only electronic database on all MoE employees. It’s based on the monthly submission of the PEDs payroll registry, and updated monthly. It includes the employees name, Bast, STEP, position, location, and duration of service. Key limitations of the database are: (I) the system is developed for salary processing, not reporting, and only maintains a single current set of information. All historic information is removed from the database; (II) the payroll registration unit was unable to generate information on haquzama inside tashkeel teachers; and (III) the database lack details on employees. EMIS – contains several education indicators including on teachers (location, gender, qualification). EMIS data on teachers (as reported by the EMIS unit) is actually based on the payroll registration database. It does not use a separate source of information. Because the EMIS unit saves payroll data annually, there is a historical record of teachers (payroll does not save historical data). However, EMIS does not report comprehensive payroll data, as it does not include the information available on outside haquzama teachers. In addition, the EMIS unit has its own source of data (the EMIS forms from schools). However, it indicates that it does not trust the reliability of the teachers data submitted by schools, see Table 9 for a comparison of payroll and EMIS self-reported date. TED dataset also known as “TEMIS� – The teacher education directors conducted a teacher survey in 2016, collecting information on teacher qualifications and compared these to the subjects they taught. It’s a dataset (based on a one- off survey) rather than a database, maintained periodically. The dataset is the only information on teacher subject qualifications. Information in the TED survey deviates somewhat from payroll data. GDHR Employee Records – The general directorate of human resource maintains two (hardcopy) HR databases. Both databases are based on submissions from the PED HR departments. The PEDs create two hardcopy employee registration files (ID, education certificates, basic information, letter of appointment) and send one to Kabul. In addition, the PEDs submit employee history/experience files (appraisals, training, past work experience). Both databases combined include much more detailed information on teachers than any other database. However, a key constraint is that the GDHR has been unable to create an electronic database (data entry of almost 300,000 files has started, but has not been completed and the process is stalled at the moment). Furthermore, the databases are outdated, many files of deceased teachers are still in there and there is no periodic updating of all files. In addition, the GDHR maintains electronic datasets on MoE attrition. However, this data, based on PED submission, is typically incomplete due to a lack of reporting compliance of PEDs. Data Gaps: There are currently several significant gaps in the information available on teachers in Afghanistan. First, there is no accurate, up to date and comprehensive information on temporary teacher contracts (haquzama). Haquzama teachers are a sizeable part of the teacher workforce, and a lack of data on this group is a serious impediment to improving teacher management. Second, there is no accurate and comprehensive information on teacher vacancies. While this information is available at PEDs, it is not structurally and periodically collected at the central level. Third, there is no information on teacher recruitment, including the number of teachers recruited each year per province, or the number of applications received for teacher positions. The latest round of teacher recruitment led by the IARCSC, will create the first detailed and comprehensive database on teacher recruitment, and an important source fro future teacher policy development. Last, due to a lack of data integration, it is currently impossible to report or produce detailed, accurate, and up to data information on teachers. Data Integration: The need to improve databases and integrate different MoE datasets has been widely accepted for several years. However, until now, the MoE has made little concrete progress. In 2017, a firm was hired to develop a single application that could absorb all databases into a single MoE HRMIS. The firm completed the first version of the application in august 2017. According to the firm, the application is ‘ready for use’ (and only require a data entry and cleaning effort of the different MoE units). The application is able (with PEDs maintenance) to produce monthly updated, detailed information on all employees of the MoE, and integrate the information available in the payroll, EMIS and GDHR database. If further developed, the application could also store biometric data, payment and overtime data, and produce automated salary payment forms. However, lack of political will, and lack of cooperation between the different units as a important impediment to implementation. Furthermore, the IARCSC announced in 2017 that it will develop a government wide HRMIS that will be applied mandatory to all line ministries. This has further slowed the process of data integration, as the IARCSC has yet to start the development of the HRMIS, and the future of the standalone MoE application is uncertain. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 49 2.6 Recommendations Recommendations – Teacher Statistics The Ministry of Education should: SHORT TERM: 134. Immediately clarify existing ambiguities regarding temporary (haquzama) teachers, including numbers and distributions across provinces, districts and schools. Information on haquzama contracts should be reported systematically and periodically verified. Haquzama data should include gender, age, qualifications and school (work location) similar to information held in regular tashkeel employee profiles. Temporary (haquzama) teachers need be part of any report or presentation by the Ministry on its workforce. Data collection on haquzama teachers should also include specification on the contract’s FTEs. Clarify Ministry employee categories. Separate teaching and non-teaching employees into different categories; identify hours worked per day/week/month and hours in front of class. Create separate categories for school support staff and MoE/PED/DED support staff. 135. Develop a clear strategy to improve the teacher GPI. Identify feasible annual targets nationally and for each province. Assess the key impediments to female recruitment in provinces lagging behind in GPI improvements, and those where improvements have been only marginal in the past years. Identify which tools can be used to promote female recruitment, and develop a comprehensive policy that enables to use of these tools in target provinces. 136. Periodically apply standardized in-service tests to all teachers, and identify teachers, subject qualifications, and regions that should be prioritized for in-service training. Standardized tests (preferably applied to all teachers periodically) will help identify key priority areas and subjects that require a focus for in-service teacher training. Address the issue of severely under qualified teachers. Identify minimum qualification standards for in-service teachers (e.g. reading, writing, basic mathematically equations), test all in-service teachers, and terminate tashkeel contracts of those in-service teachers that do not meet the minimum standards. An appropriate policy needs to be developed to suggest termination, in-service training, or continuation based on test results, and to guarantee rigorous oversight of standardized in-service tests. Furthermore, minimum qualification standards should be specified for teachers in different grades and school types (primary, lower and higher secondary). 137. Develop a comprehensive, multi-year plan to address pre-service teacher training needs by province, district, gender, and subject specialization. This plan should be based on existing vacancies (Identify a framework for school staffing based on the curriculum and student to teacher norms, in order to identify the needs for subjects by FTE), and projected teacher recruitment and compare the number of vacant positions to the number of available qualified and willing applicants (from the IARCSC database). The plan should identify districts with severe shortages of subject-qualified candidates (by gender) to be prioritized for teacher training services. 138. Improve data collection on teacher attrition (gender, age, location, qualification, reason for leaving), teacher vacancies (location, gender, subject), and applications, in order to utilize this data for teacher policy and management improvements. 139. Develop a comprehensive, up-to-date, accurate electronic database of all MoE managed Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 50 teachers (and other employees). Integrate existing data (GDHR, EMIS, Payroll) into a single HR application that enables quick information extraction, administration of HR related management files and macro (national) and provincial level reporting and analyses. The integrated system should include employees profiles, hours worked, leave days and salary payments, for permanent and temporary employees. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 51 3 Teacher Needs and Allocation 140. This chapter discusses the norms and methods used by the Ministry of Education to determined teacher needs and the allocation of new teacher positions. Norms and methods are determined to identify how many teachers there should be, and where. It thereafter describes how tashkeel teachers are actually distributed across provinces in Afghanistan, how they have historically been allocated across provinces, and then provides an initial look at teacher distributions across schools. This chapter ends with a discussion and recommendations. 3.1 MoE Norms on Teachers Needs 141. The Ministry of Education’s norms, used to calculate the number of teachers (and other staff) needed for schools, are described in a policy document called: Educational Institution’s Planning Standards (2007), see Annex 1. The policy document prescribes different norms for teachers and support staff (academic and non-academic) for different types of schools included in the MoE’s five main programs (General Education, Islamic Education, Teacher Education, Technical and Vocational Educational and Training, or TVET, and Literacy). 142. The planning standards use a class-based method to identify teacher needs. For general education (responsible for 95% of all students attending MoE classes), the standards identify the number of students per class for grades 1 to 12, and the number of teachers per class from grades 1 to 12. For example for grades 10 to 12 the norm is 30 students per class and 1.5 teachers per class. Although not clearly indicated in the document, the number of teachers per class appears to indicate the number of teaching hours (i.e. 1.5*24 teaching hours = 36 teaching hours per week in grades 7 to 12). Please see Figure 12 for class-based norms. Figure 12: MoE student- and teacher- norms per class 1.4 1.5 1.5 40 40 35 30 1 Grade 1-3 Grade 4-6 Grade 7-9 Grade 10-12 Grade 1-3 Grade 4-6 Grade 7-9 Grade 10-12 Students per Class Teachers per Class (Source: MoE Educational Planning Standards 2007) For the purpose of analyses later, these class-based norms have been recalculated to pupil- teacher ratios (PTRs) see Figure 13. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 52 Figure 13: Converted PTR norms and assumed teaching hours per grade 33.6 36 36 40 29 24 23 20 Grade 1-3 Grade 4-6 Grade 7-9 Grade 10-12 Grade 1-3 Grade 4-6 Grade 7-9 Grade 10-12 Pupil-Teacher Ratio (converted) Teaching Hours per Class (assumed) (Source: Staff calculations) 143. Figure 14 provides an overview of actual regional PTRs in secondary education, compared to Afghanistan’s norms for secondary education. A norm between 23 and 20 for secondary education fits trends, both regionally and compared to other low-income countries. Figure 14: Comparison of actual regional secondary education PTRs to Afghanistan's norms 32 27 23 23 20 21 18 13 Afghanistan Afghanistan India Pakistan Iran Central and Low Income High Income Grades 7-9 Grades 10-12 South Asia Countries Countries (Source: UNESCO, 2016) 144. In addition, the standards identify norms for academic, managerial and non-academic support staff based on the number of students per school or the availability of facilities. For example, schools with an auditorium should have an audiovisual teacher; schools with over 2,500 books should have a librarian teacher and if the number exceeds 4,000 there should be a librarian assistant as well. Other criteria are identified for drivers (if the school has a functioning car), cleaners (one per six classes even if there are no classrooms), typists, maintenance officers, technical workers, gardeners, cashiers, work managers (managing support staff), assistant teachers (one per 2,000 students), heads of departments, and lead teachers for chemistry, biology, physics and math laboratories. 145. Similarly, the standards identify norms for dormitories (dormitory teachers, Mullah Imams, tailors, barbers, cooks, tea makers, cashiers, dishwashers, cleaners, and laundry workers), and other MoE programs (TVET, Islamic Education, TTCs, and Literacy). The standards also indicate that for each 10 female teachers, there should be an assistant teacher to cover for maternity leave. 146. There are four main issues that limit the practical application of these planning standards for planning purposes. First, the existing norms on staffing schools are aspirational rather than Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 53 realistic. If fully applied, the MoE would have to double its workforce. The institutional standards describe the staffing of a school in an ideal situation without human resources or fiscal constraints. Written in 2006, the norms were developed in a period when Afghanistan’s long-term fiscal outlook was much more opaque. Moreover, since then, key indicators in the education sector (e.g. number of schools, students, and teachers) have significantly changed. If the standards would be literally applied, all schools would require a significant increase in academic and non-academic support staff. Tailors and barbers for dormitories, or cleaners for every six classrooms might now seem unjustifiable given the government’s aim to contain the civil servant wage bill. 147. Second, the norms are not comprehensive, not detailed, and not consistent. The ‘educational institution’s planning standards’ document is subdivided in sections by the main five MoE programs (GE, IE, TVET, TTC, and Literacy). For example, the standards identify the number of teaching hours for TVET lecturers, but not for general education teachers. In the ‘general education’ section, minimum standards are identified to create a primary school (i.e. two classes with at least 25 students), but no such standards are given for secondary and high schools. It is also not clear from the document what are the minimum and maximum class sizes in general education35. For example, what should a school with 42 grade 1 students do? Create one class (with 42 students) or two classes (with 21 students each). And what if the school has 55 grade 1 students? The standards also do not specify subject specialization of any of the teachers (e.g. at what point or size should a school have at least one specialized chemistry teacher?). Importantly, all MoE officials interviewed for this study indicated that a minimum of ten enrolled students is required to create a class in general education. 148. Third, the norms do not identify a method to prioritize teachers (and school staff) needs under conditions of scarcity. Given the aspirational nature of the standards, and systemic shortages of staffing at schools across the country, how should schools determine what they need most? If a school has twelve vacant positions, determined by the standards, than which should be filled first? Given that not all will be staffed. There are also no instructions on how PEDs, or DEDs should prioritize new teacher positions across schools. Should a teacher be allocated to a school with 95 students in a class (and only one teacher), or to a school with 12 students in a class (but not a single teacher)? 149. Last, and importantly, MoE staff involved in the assessment of teacher needs do not accurately know the needs assessment norms. All MoE staff interviewed for this study, including teachers, school principals, PED staff and officials of many different directorates within the central ministry, were asked to reproduce the teacher needs assessment norms. Whereas all officials used similar methods to calculate teacher needs (i.e. a class-based ratio, with progressively declining numbers of students in higher grades), they all described somewhat deviating norms. Typically, officials would prescribe 1.25 teachers per class in grades 4 to 6, or 7 to 9, or indicate a lower number of students per class in grades 4 to 6. While these deviations might appear minor, they significantly alter the number of teachers needed in larger schools, provinces, and nation-wide (with more than 9 million students enrolled). 35 In the Educational Institution’s Planning Standards there is a section that describes class sizes (page 5 translated version): for elementary 10 to 30, secondary and high schools 10 to 20, and grades 13-14 also 10 to 20. However, this falls under the header of ‘Technical Institutions’ (TVET). It is unclear if these margins are meant for general education, especially since they do not correspond to other norms given in the document (i.e. 40 student per class in grades 1-6). Moreover, the minimum standard of 10 enrolled students (i.e. 7 to 8 attending students) allow for the creation of classes with PTRs less than half of the norm in high income countries. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 54 3.2 Needs assessments in practice 150. In practice, the norms applied to identify teacher needs vary considerably. It part this seems to be the result of a lack of awareness amongst planners. Few officials have a copy of the institutional standards, and none of the officials interviewed was fully aware of the details in the regulation. In part this might be caused by the ambiguity of the norms. And in part this could be caused by the sustained shortage (as defined by the standards) of human resources, where those responsible for assessing needs have adapted their methods to reflect the anticipated availability of HR. 151. As a result, two school principals, with similar numbers of students and teachers, might conclude they need significantly different numbers of additional teachers, based on the application of different norms. School principals and PED officials, interviewed for this study, indicated using ‘their own’ norms, rather than those prescribed by the MoE standards. One school principal indicated that all classes (grades 1 to 12) had around 46 students enrolled. PED officials also reported using deviating norms in order to plan given actual limitations on the number of teachers they are allocated. 152. The MoE uses two different processes to assess teacher needs. The first is an internal consultative process where the central Ministry asks its provincial directorates to provide overviews of teachers needed. This request trickles down to schools where teacher needs are calculated (usually by school principals or headmasters, occasionally with the support of DED staff). This information is send upward, and collated at district-, provincial, and eventually the central (national) level. The reliability of this process is problematic, as there seem to be a multitude of different norms and calculation methods applied by officials assessing teacher needs. In the second process, the MoE general directorate of planning and policy (GD P&P) uses its database on students, classes and schools (EMIS) and teachers (payroll registration database) to determine the number of teachers needed across the country. 153. However, methods for calculating teacher needs at the central MoE are also not standardized. The GD P&P shared multiple teacher needs assessments for this study, in all containing four different calculation methods. The methods use different indicators for calculating the number of students, i.e. enrolled or attending students (usually 70% to 80% of enrolled students). The methods also apply different factors for multiplication and/or additions (i.e. for calculating support staff). And importantly, the methods diverge in their indicator on class size, either using ‘actual class size’ (based on EMIS data) or ‘adjusted class size’ (assuming classes have the average size of the student-per-class norm)36. Moreover, none of the methods used by the GD P&P considers the numbers of haquzama teachers (inside and outside tashkeel). Not surprisingly, the methods have significantly different outcomes (i.e. between 31,000 and 91,000 additional staff needed37). 154. Please find below two examples of the teacher needs calculation prepared by the MoE GD P&P in 2017. The first calculation method (Table 5) uses the actual number of classes per grade (based on EMIS data) and multiplies them by the teacher-per-class norm38. The second method 36 Note that when a norm is applied on the average student class size, rather than actual class sizes, the method (teacher norm combined with average student norm) is essentially the same as applying pupil-teacher ratios to calculate teacher needs. 37 Data retrieved from actual MoE GD P&P calculations. 38 Note that the MoE applied a teacher-to-class ratio of 1.25 for grades 4 to 6 in this method, instead of the 1.4 prescribed by the institutional standards. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 55 (Table 6) uses the number of attending students and divides them by the student-to-class norm. As such, the second method adjusts the number of classes, based on that the average class size should correspond to the norm. The second method is the equivalent of applying PTRs to calculate teacher needs. Table 5: 'Actual Class Size" method for calculating teacher needs Students Existing # of Students Teachers Teachers Grades (attending) Classes per Class per Class Needed Grade 1-3 2.494.683 73.271 34 1,00 73.271 Grade 4-6 2.147.123 66.917 32 1,25 83.646 Grade 7-9 1.357.671 44.214 31 1,50 66.321 Grade 10-12 788.198 26.693 30 1,50 40.040 Grade 13-14 56.975 2.274 25 1,50 3.411 Total 6.844.650 213.369 266.689 (Source: MoE GD P&P 2017, amended) 155. Using the ‘actual class size’ method, this calculation concludes that there are 266,689 teacher required in total, or 63,048 additional teachers on top of the 203,641 existing tashkeel teachers. In addition the MoE GD P&P adds that 10% additional female assistant teachers should be hired (to cover for maternity leave) and 25% school support staff, putting the number of total required additional staff at 87,210.39 Table 6: 'Adjusted Class Size" method for calculating teacher needs Students Adjusted # of Students per Teachers Teachers Grades (attending) Classes Class per Class Needed Grade 1-3 2.494.683 62.367 40 1,00 62.367 Grade 4-6 2.147.123 53.678 40 1,25 67.098 Grade 7-9 1.357.671 38.791 35 1,50 58.186 Grade 10-12 788.198 26.273 30 1,50 39.410 Grade 13-14 56.975 1.899 30 1,50 2.849 Total 6.844.650 183.008 229.909 (Source MoE GD P&P 2017, amended) 156. Using the ‘adjusted class size’ method, this calculation concludes that there are 229,909 teacher required in total, or 26,268 additional teachers on top of the 203,641 existing tashkeel teachers. Similarly, the method adds female assistant teachers and school support staff, putting the number of total required additional staff at 41,235. 39This calculation adds 10% female assistant teachers over all additional teachers required (rather than just the female teachers in this pool) and thereafter multiplies the total (110%) by another 25% support staff. The actual calculation of the MoE DG P&P deviates somewhat from the one presented above, as it uses 2016 payroll data to identify existing teacher number, rather than 2017 data. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 56 157. Note that there are already an estimated 35,824 teacher FTEs through haquzama contracts, which are not considered in these calculations. In addition, the MoE GD P&P does did not use student numbers (enrolled and attending) corresponding to EMIS data, and the actual number of students enrolled is somewhat higher (according to EMIS). 158. Using the ‘actual class size’ method makes sense given the significant discrepancies between class sizes across different schools. The policy of allowing the creation of classes at a minimum of ten enrolled students, in order to promote access to education in rural areas, has resulted in a very large number of schools with class sizes lower than the norm (two-thirds of almost 6,000 primary schools reported through EMIS having on average less than 40 students per grade). 159. However, the use of the ‘actual class size’ calculation method (rather than the ‘adjusted class size’ method or applying converted pupil-teacher ratios) is also problematic, for it means that teacher needs are effectively determined by class size. The more classes with few students there are, the more teachers are needed, even with the same amount of total students. Schools, DEDs, and PEDs are essentially incentivized to create many small classes as a way to compete over pubic resources (i.e. teachers). This challenges teacher planning, as there are no clear criteria for, and controls on the creation of classes (beside the widely reported minimum of ten enrolled students). Conversely, principals interviewed for this study appeared to anticipate HR constraints and did not create additional classes (even though student numbers exceeded 60) because they thought they would not get the necessary teachers for it. Using the ‘actual class size’ requires better (and tighter) regulation of class size, as well as significantly more resources (i.e. teachers). If not, it risks becoming a first come first served system. 160. Moreover, the ‘actual class size’ method to calculate teacher needs, requires detailed knowledge of the size of each individual class (and there are more than 200,000 classes). The needs of teachers per province, or nation-wide cannot be calculated using the total number of students (per grade), but instead requires a complex calculation using detailed datasets on the composition of classes. This prevents national, subnational, and international education stakeholders (NGOs, provincial education directors, governors, council members, MPs, Ministry of Finance officials, the President’s Office, and international development partners) from externally validating the teacher needs assessments. Provincial Council members are currently unable to determine if their province has received an equitable amount of teachers, based on the number of students in the provinces alone. Instead, they require access to detailed EMIS datasets, and the capabilities to analyze them. 161. It is important to stress here that the Ministry’s class-based norms or targets for pupil- teacher-ratios (PTRs) are not clearly identified in any policy or strategic document, besides the Educational Institution’s Planning Standards. The MoE has not identified target PTRs per province or nation-wide, and neither has it published its methods for calculating teacher needs. As a result it is unclear to what extent these targets are based on consensus within the Ministry and the Afghan Government, or if these targets have politically balanced the need for quality education against fiscal constraints. 162. Last, if the ‘actual class size’ method is used for planning purposes (i.e. identifying where teachers should be allocated), and there is no average norm for a pool of classes (average PTR norm for schools, districts or provinces), it can result in extreme variations between the numbers of students per teacher, in classes across the country. This issue will be described in more detail in section 3.4 teacher allocation. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 57 3.3 Teacher Needs Forecasting 163. Teacher needs forecasting is challenging because it is difficult to predict future student numbers without reliable indicators. There is no accurate headcount on the numbers of school- aged children in Afghanistan. Afghanistan’s last census was in 1979. Since then, public institutions have used estimates of (school-aged) population. Net and gross enrolment rates are therefore ‘uncertain’ estimates. However, UNICEF estimated 3.5 million children were out of school in 2017 (girls comprise 75% of this group)40 and if the MoE were to effectively address this issue, student enrollment, as well as teachers needed, would rise more rapidly. Similarly, there are only uncertain estimates of population growth. 164. The EMIS unit maintains the only database on enrolled student numbers. EMIS student data is based on school self-reporting, and the reliability of EMIS student numbers is often disputed41. EMIS data shows an increase of 30.3% in enrolled students between 2010 and 2017 (approximately 3.9% year-on-year). However, yearly growth of student enrollment has plateaued somewhat in the last three years (between 1.5% and 2.2% year-on-year). It is unclear if this trend will continue. Figure 15: Number of enrolled students 2010-2016 and annual growth in percentage 9,234,459 9,389,901 8,901,588 9,034,403 8,481,230 8,018,255 7,507,564 7,205,285 N/A +4.2% +6.8% +5.8% +5.0% +1.5% +2.2% +1.7% 1389 1390 1391 1392 1393 1394 1395 1396 2010 2011 2012 2013 2014 2015 2016 2017 (Source: EMIS) 165. In this section, two scenarios on teacher needs are presented using different norms and methods to calculate teacher needs. The scenarios both share the following assumptions: I. The number of enrolled students will continue to increase by 3.9% year-on-year; II. The percentage of attending students will remain unchanged (around 76.4%) until 2025; III. The relative distribution of students across grades will remain unchanged until 2025; 166. Applying these assumptions, the number of enrolled students will increase from almost 9.4 million42 in 2017 to a little over 12.7 million in 2025. The number of attending students will increase by 35% from 7.2 million in 2017 to 9.7 million in 2025, see Figure 16. The number of attending students will be used as the primary driver to assess future teacher needs. 40 UNICEF Afghanistan Annual Report 2017, p3. UNICEF 2017. 41 EMIS data verification report, p8. Afghanistan Institute of Rural Development & Afghanistan Holding Group 2016. 42 Note that the number of enrolled students in EMIS data is somewhat higher than the number of enrolled students used in the GD P&P teacher needs calculation. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 58 Figure 16: Projected number of attending and enrolled students 2017-2025 12,654,350 9,389,901 9,316,864 9,670,905 8,647,191 8,975,784 8,025,652 8,330,627 7,448,788 7,731,842 7,176,097 1396 1397 1398 1399 1400 1401 1402 1403 1404 2017 2018 2019 2020 2021 2022 2023 2024 2025 Present Enrolled (Source: Staff Calculations) 167. Furthermore an average PTR (grades 1 to 14) is calculated for the different methods used to assess teacher needs. This is done by dividing the number of students by the outcome of the teacher needs method. If the relative size of grades remains unchanged until 2025, the average PTR (grades 1 to 14) remains unchanged. The ‘actual class size’ method has an average PTR of 25.7. The number of teachers needed using this method, increases form 279,226 in 2017 to 376,300 in 2025, see Figure 17. The ‘adjusted class size’ method has an average PTR of 29.8. The number of teachers needed using the ‘adjusted class size’ method, increases form 240,809 in 2017 to 324,527 in 2025. Note that the actual net PTR (based on estimates of FTEs and attending students) is 30, only marginally higher than the adjusted class size PTR, and the estimated number of teacher FTEs, 239,456, is only marginally lower than the calculated need by this method. Using the ‘adjusted class size’ method to calculate future teacher needs is practically the same as calculating future teacher needs if the current PTR would be maintained. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 59 Figure 17: Forecasted teacher needs 2017-2025 using actual and adjusted class size methods 376,300 362,524 349,252 336,467 324,149 324,527 312,282 312,646 300,850 301,201 289,836 290,174 279,226 279,551 269,317 259,458 249,959 240,809 1396 1397 1398 1399 1400 1401 1402 1403 1404 2017 2018 2019 2020 2021 2022 2023 2024 2025 Actual Class Size' Method (PTR 25.7) Adjusted Class Size' Method (PTR 29.8) (Source: Staff Calculations) 168. The number of (estimated) teacher FTEs in 2017 is only marginally lower than the teachers needed using the ‘adjusted class size’ method, but is far less than the needs of the ‘actual class size’ method. In order to consider the current number of teachers, the two scenarios are adjusted to progressively address the teacher gap until 2025, see Figure 18. Figure 18: Forecasted teacher needs 2017-2025 adjusted to progressively adress teacher gap 376,300 355,630 336,093 324,527 317,630 312,423 300,180 300,774 289,560 283,689 278,763 268,104 268,369 253,376 258,363 248,730 239,456 1396 1397 1398 1399 1400 1401 1402 1403 1404 2017 2018 2019 2020 2021 2022 2023 2024 2025 Actual Class Size' Method (PTR 25.7) Adjusted Class Size' Method (PTR 29.8) (Source: Staff Calculations) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 60 169. In the lower estimate (blue line, adjusted class size method), a 3.87% year-on-year teacher growth is required to adjust for student growth (and address the marginal gap). In 2018, an additional 9,274 teachers will be required, and this number will progressively increase to 12,100 additional teachers in 2025, see Figure 19. Using the ‘adjusted class size’ method, the total number of teacher FTEs needs, will increase by 85,071 (36%) until 2025. 170. In the higher estimate (orange line, using the ‘actual class size’ method) a 5.81% year-on- year teacher growth is required to adjust for student growth and address the existing teacher gap. In 2018, an additional 13,920 teacher will be required, and this number will progressively increase to 20,673 additional teachers in 2025. Using the ‘actual class size’ method, the total number of teacher FTEs needs, will increase by 136,844 (57%) until 2025. 171. Note that in both forecasting scenarios the number of additional teacher FTEs required each year is considerable, between 9,274 and 13,920 in 2018, and between 12,100 and 20,673 in 2025. The hiring of an additional 8,000 tashkeel teachers (ongoing in early 2018), if combined with similar growth in haquzama FTEs (an additional 15% to 20%) is only sufficient to sustain the current PTR. And if student enrollment continues to increase as it did in the past seven years, the MoE will need hire more than ten thousand teachers each year, simply to sustain the current PTR. The forecasted number of teachers needs presented in Figure 18, will be used in chapter 0 to forecast teacher and MoE wage bill expenditures. Figure 19: Annual number of additional teacher FTEs needed 2018-2025 using two different calculation methods 20,673 19,537 18,464 17,449 16,491 15,585 14,729 13,920 11,649 12,100 10,797 11,215 10,006 10,394 9,274 9,633 1396 1397 1398 1399 1400 1401 1402 1403 1404 2017 2018 2019 2020 2021 2022 2023 2024 2025 Adjusted Class Size' Method (PTR 29.8) Actual Class Size' Method (PTR 25.7) (Source: Staff Calculations) 3.4 Teacher Allocation 172. This section will describe how teachers are allocated across provinces in Afghanistan, how this has evolved, and will look at the distribution of teachers across schools. Allocations will be described in terms of pupil to teacher ratios (PTRs). As noted in chapter 89, PTRs significantly diverge depending on the definition and indicators used. There are on average 46 enrolled students per tashkeel teacher in Afghanistan, but only 39 attending students. Furthermore, if total teacher FTEs are considered (including haquzama teachers), these numbers decrease to 35, and 30 respectively. Unless otherwise indicated, this section describes the enrolled student to tashkeel teacher ratio (46 on average), because MoE databases are the most comprehensive, detailed and reliable on these indicators. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 61 3.4.1 Provincial Teacher Allocation 173. In 2017, some provinces had twice as many tashkeel teachers as other provinces, for the same number of enrolled students. Nuristan, Panjshir, Badakhshan, Wardak, Parwan, and Baghlan each have less than 36 students per teacher, whereas Helmand, Kandahar and Khost have between 60 and 72 students per teacher, see Map 4. The significant differences in these PTRs cannot be explained by either the geographic composition of these provinces (i.e. more small classes in rural schools across large mountainous areas), or the relative size of grades (i.e. the age composition of students). Either these discrepancies are the result of the ‘actual class size’ method for calculating teacher needs (and some provinces are more successful in creating new classes, even though they don’t have more rural villages/communities), or the allocation of teacher tashkeel positions has not been based on the student needs reflected in the data. Map 4: Provincial PTRs (number of enrolled students per tashkeel teacher) 2017 43 46 50 33 44 42 44 35 46 31 31 50 39 35 42 42 42 58 34 52 51 41 44 52 50 55 72 42 40 42 48 KABUL CITY 50 46 NATIONAL AVERAGE 60 61 48 (Source: Payroll Registration Database, EMIS) 174. Furthermore, in the past seven years, the significant variations in provincial PTRs have not been decreased through the allocation of new tashkeel teacher positions 43 . Some provinces (such as Panjshir, Badakhshan, Wardak and Parwan) with the lowest PTRs in 2011 received significant additional teachers, whereas some provinces (Herat and Daikundi) with the highest PTRs received relatively few additional positions, see Table 7. And, the average PTR gap between provinces did not decrease in the past 7 years (i.e. the provincial PTR standard deviation increased between 2011 and 2017). 43An annual standard deviation was calculated as a measure of variation between provincial PTRs (ST.DEV). This indicator increased from under 6 in 1390 to 9 in 1396. However, ST.DEV appears to be affected by extreme PTR outliers likely caused by errors in EMIS data submission or entry. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 62 Table 7: Pupil-Teacher Ratios between 2011-2017 (* some data unreliable) Province 2011 2012 2013 2014 2015 2016 2017 +/- Kabul City 39 40 39 41 41 45 48 +9 17 provinces with Lowest PTR 35 37 38 40 41 42 42 +7 Badakhshan 30 34 36 33 34 34 33 +3 Panshir 30 35 32 36 33 30 31 +1 Lowest Pupil- Parwan 31 31 31 32 32 34 35 +4 Teacher Ratios Paktika 31 38 40 47 46 44 40 +9 Baghlan 31 34 34 34 33 34 35 +4 Wardak 31 32 35 35 36 34 34 +3 Provincial Average (ex Kabul City) 41 42 42 44 45 46 46 +5 17 provinces with highest PTR 46 46 46 48 48 50 49 +3 Herat 46 47 47 45 48 53 51 +5 Laghman 46 45 44 42 42 42 42 -4 Highest Pupil- Uruzgan 47 51 49 45 42 47 42 -5 Teacher Ratios Daikundi 49 51 66* 89* 52 53 52 +3 Nangarhar 50 51 49 50 51 50 52 +2 Badghis 101* 51 53 59 52 53 50 -1 (Source: EMIS, Payroll Registration Database) 175. Moreover, in several provinces the number of students rose dramatically between 2010 and 2016. According to EMIS data, Kandahar, Kabul Province, Khost and Helmand saw the number of enrolled students almost double in this period (+76%, 81%, +94%, and +106% respectively), yet none of these provinces received significantly more teachers than the national average. As a result, in these provinces the number of students per teacher rose considerably, and in 2017, their PTRs are almost double that of provinces with low PTRs (Panjshir or Wardak) see Table 8. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 63 Table 8: Provinces with rapid increase in PTR 2011-2017 Students Teachers PTRs Province 2010 2016 +/-% 2011 2017 +/-% 2011 2017 Panshir 35.929 41.983 + 17% 1.192 1.347 + 13% 30 31 Wardak 131.080 158.471 + 21% 4.456 4.613 + 4% 31 34 Kandahar 133.306 235.247 + 76% 3.961 4.156 + 5% 38 61 Kabul Province 177.345 321.384 + 81% 5.190 5.779 + 11% 38 58 Khost 161.314 313.593 + 94% 4.112 4.441 + 8% 39 72 Helmand 88.887 182.990 +106% 2.465 2.780 + 13% 43 60 (Source: EMIS, Payroll Registration Database) 176. There are justifiable reasons to differentiate provincial PTR targets/norms. For example, less urbanized provinces might need smaller schools (with lower PTRs) to provide access to rural communities. However, the PTR gaps shown above cannot be explained from an urban/rural divide, nor would it justify the size (i.e. some provinces having twice as many teachers as others). Furthermore, the MoE does not have any written policy on PTR norms, nor on differentiating teacher needs based on degree of ‘urbanization’/concentration. 177. Most MoE officials interviewed for this study did not know how provincial teacher allocations were determined, which department was responsible for this, or which officials approved provincial distributions 44 . Some interviewees confided that they perceived the allocation process to be ‘political’ rather than ‘needs-based’. There is also no database or historical record of how new teacher positions have been allocated in the past years. Note that the above presented data reflects tashkeel teachers employed (and registered in the payroll database), and not the number of allocated tashkeel positions (i.e. approved number of staff including vacancies). 178. In late 2017, the MoE allocated 8,000 new tashkeel teacher positions across its 35 PEDs. This list of vacancies was created to inform the recruitment of new teachers led by the IARCSC. The new positions are not exclusively allocated to provinces with high PTRs. All provinces received additional teacher positions, including those that already had very low PTRs (Parwan, Wardak, Panjshir, Badakshan and Baghlan), and the provincial PTR gap has not been reduced by this distribution45. However, the relative size of new teachers allocated to provinces closely correlates with the relative size of teacher needs identified by the ‘actual class size’ method (r=.81). The distribution of new positions also correlates with the number of existing teachers in provinces (r=.61). This seems to suggest that the recent allocation of 8,000 new teachers is mostly based on a teacher needs assessment using the ‘actual class size’ method. 44 One MoE official indicated that in past years there was a committee that established the provincial teacher allocation each year, led by the senior MoE leadership. 45 Standard deviation for the old and new provincial PTRs did not significantly change (old standard deviation provincial PTRs= 8.99, new= 8.94). Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 64 3.4.2 Teacher Allocation Across Schools 179. Data on the distribution of teachers across schools is less reliable. There is no school- disaggregated data available in the payroll registration database (widely perceived as the most reliable database on teacher numbers)46. The EMIS unit maintains a database disaggregated at school level (including student and teacher numbers), however, the unit cautions the use of school level data for lack of reliability. EMIS data is based on self-reported data by schools. Schools fill in a fairly complex 8-page form, which includes information on teachers and students. Out of the 17,861 schools registered in the database, approximately 92% submitted data to EMIS. 180. The database contains many inconsistencies and errors, however an analyses of aggregated sums (teachers per province), presented in Table 9 shows that EMIS teacher data does correspond to payroll registration data47, and the size of the database allows for analyzing macro-trends across a large pool of schools. The following analyses will focus on primary schools only. There are 6,695 primary schools registered in the EMIS database, of which 5,715 (85%) submitted reliable data48. Within this database, there is one teacher for every 53 students enrolled in primary education. 46 School names are identified in the payroll profile, but the database does not allow for calculating teachers per school (as schools names are not categorized). 47 The EMIS school-level dataset reports a total of 9,383,935 students and a total of 216,447 teachers. Compared to 203,641 tashkeel teachers, 14,620 haquzama outside tashkeel, (and 21,204 haquzama inside tashkeel) in the payroll registration database. Furthermore, EMIS reports 10,524 teachers in Badakhshan, whereas 10,503 are registered in the payroll database. In addition to data submission errors (by the schools) and data entry errors (by the EMIS unit) discrepancies might be caused by number of closed schools (not submitting data but still having teachers on payroll) or the confusion over haquzama contracts (the payroll registration database reports the MoE employs 218,261 people). The most significant outlier, is the EMIS data on teachers in Kabul City, where more than 10,000 teachers were reported by EMIS. 48 This data pool is chosen because primary schools have only 6 grades with similar student and teacher to class ration, whereas secondary and high schools often also include lower grades creating more variance. There are 1,345,671 students in this pool, and 25,258 teachers according to EMIS. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 65 Table 9: Comparison of Payroll registration and EMIS data on teachers per provinces 2017 Payroll Registration Database EMIS Database Difference Male Female Total Male Female Total % Kabul Prov 4.076 1.703 5.779 4.041 1.756 5.797 0% Kapisa 2.962 591 3.553 2.915 565 3.480 -2% Parwan 5.254 963 6.217 5.169 960 6.129 -1% Wardak 4.347 266 4.613 4.453 256 4.709 2% Logar 2.814 446 3.260 2.868 351 3.219 -1% Ghazni 4.948 1.778 6.726 5.794 1.773 7.567 13% Paktia 3.557 245 3.802 3.393 176 3.569 -6% Nangarhar 12.805 1.604 14.409 12.589 1.998 14.587 1% Laghman 4.181 325 4.506 3.830 357 4.187 -7% Kunar 3.884 252 4.136 3.648 221 3.869 -6% Badakhshan 6.941 3.562 10.503 7.128 3.396 10.524 0% Takhar 5.618 2.828 8.446 6.338 2.591 8.929 6% Baghlan 6.888 2.491 9.379 6.472 2.554 9.026 -4% Kunduz 4.816 2.107 6.923 4.703 2.153 6.856 -1% Samangan 1.828 782 2.610 1.743 661 2.404 -8% Balkh 5.134 6.958 12.092 6.037 7.557 13.594 12% Jawzjan 2.610 2.034 4.644 2.492 1.999 4.491 -3% Faryab 4.615 2.805 7.420 3.957 2.226 6.183 -17% Badghis 2.049 518 2.567 1.914 489 2.403 -6% Hirat 7.597 8.248 15.845 9.145 9.261 18.406 16% Farah 1.745 1.001 2.746 1.736 977 2.713 -1% Nimroz 610 826 1.436 708 919 1.627 13% Hilmand 2.281 499 2.780 2.147 538 2.685 -3% Kandahar 3.375 781 4.156 2.915 629 3.544 -15% Zabul 940 84 1.024 898 72 970 -5% Uruzgan 1.469 135 1.604 1.378 130 1.508 -6% Ghor 3.967 470 4.437 4.362 449 4.811 8% Bamyan 2.632 864 3.496 2.624 842 3.466 -1% Paktika 3.337 58 3.395 3.083 83 3.166 -7% Nuristan 1.306 178 1.484 1.054 110 1.164 -22% Sar i Pul 2.099 1.297 3.396 2.056 953 3.009 -11% Khost 4.253 188 4.441 5.311 189 5.500 24% Panjshir 1.044 303 1.347 961 278 1.239 -8% Daikundi 1.932 1.242 3.174 1.965 1.207 3.172 0% Kabul City 7.614 19.681 27.295 11.573 26.371 37.944 39% TOTAL 135.528 68.113 203.641 141.400 75.047 216.447 6% (Source: EMIS school level database 2017, Payroll registration Database 2017) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 66 181. There are tremendous variations in PTRs across schools. Almost 23% of primary schools (n=1,295) reported having more than 100 students per teacher, and almost 13% (n=719) reported having more than 140. Half of the schools reported having between 20 and 70 students per teacher, and 11% reported having less than 20 students per teacher, see Figure 20. It is important to consider here that these are figures for enrolled students. The number of students permanently absent usually varies between 20% to 30%. Figure 20: Number of Primary schools within PTR ranges 719 670 654 591 515 449 450 378 296 195 216 201 156 109 110 3% 8% 9% 12% 11% 10% 8% 7% 5% 4% 4% 3% 2% 2% 13% (Source: EMIS 2017) 182. While these variations might appear extreme, it is important to consider the ‘student compensation burden’ if small (rural) classes are being created. Only 10 enrolled students are required to create a “small� class (and thus a teacher position). However, with an average of around 53 enrolled students per teacher in primary education, this means that the remaining 43 students need to be added to another “large� class (thus resulting in a class of 96 enrolled students and 1 teachers) see Figure 21. With twice as many “small� schools as “large� schools, and some provinces having double the number of students per teacher, this logically results in some schools having PTRs over 150. Indeed 23% of all primary schools, that submitted data to EMIS, reported having more than 100 students per teachers. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 67 Figure 21: ‘Student Compensation Burden' illustrated with PTR 53 in primary education +70 53 53 53 -40 -30 53 23 13 school 1 school 2 school 3 school 1 school 2 school 3 even distribution large schools compensating for smaller schoools (Source: Staff Calculations) 183. Similarly, the 20% of schools reporting the highest PTRs, reported having on average only one teacher per 176 students. And the 20% of students enrolled in schools with the highest PTRs have on average only 1 teacher per 226 students. Figure 22: Average PTRs (enrolled) Primary Education per quintile of schools (n=5.715) and students (n= 1.345.671) 226 176 108 83 70 56 48 38 16 22 Lowest Lower Middle Higher Highest Lowest Lower Middle Higher Highest School Quintile Student Quintile (Source: EMIS 2017) 184. These are extreme figures and a note of caution is required. EMIS is based on ‘self- reported’ school data. It is possible that schools structurally over report the number of students, or structurally under report the number of teachers (though teacher numbers do correspond with payroll data). However, these figures cannot be attributed to individual errors. And logically, small schools create enormous classes elsewhere. 185. While more reliable (and verified) data is required to be conclusive, it is clear that there are substantial discrepancies between schools. Currently, some schools might have ten times more enrolled students per teacher. This calls into question the costs of the low benchmark for creating classes, as well as the effectiveness and efficiency of past teacher allocation methods. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 68 3.4.3 Haquzama Teachers 186. Data provided by different MoE units on the numbers of haquzama teacher per province is inconsistent and incomplete, see Table 10. The payroll unit was only able to provide numbers of haquzama outside tashkeel teachers, and these numbers varied considerably from the data provided by the budget unit (on both haquzama inside and outside tashkeel teachers). However, in both datasets, the distribution of haquzama teachers does not significantly reduce the PTR gap between provinces. Haquzama teachers are not exclusively working in provinces with high PTRs, but also in provinces with low PTRs. 187. Importantly, unlike tashkeel positions, the numbers of haquzama teachers (and their provincial distribution) are not centrally determined or approved. The numbers of tashkeel positions per line ministry are strictly regulated, and new tashkeel positions require approval form the President’s Office. The provincial distribution of new tashkeel positions is determined and approved by the MoE’s senior leadership. However, there seems to be no such restrictions on the number of haquzama contracts. PED officials interviewed for this study indicated they did not receive an indication of the maximum number (ceiling) of haquzama contracts from the central MoE. 188. The primary limitation on haquzama contracts is budget availability. Haquzama salaries are paid out of an ‘overtime’ budget line, and the size of this ‘overtime’ budget allotment is determined by the central MoE. However, one PED official indicated he could hire as many haquzama teachers as he deemed necessary, and that every year his PED overspent on the ‘overtime’ budget line (resulting in arrears that were later adjusted by central MoE budget officials). He also reported having significantly more haquzama teacher contracts in his province than registered in either database. However, some PEDs did not hire any haquzama teachers (according to both datasets), suggesting there is a variation in PED practices. Unfortunately, this study did not interview sufficient PED officials to indicate trends across the country. 189. In order to provide an indication of distribution of haquzama teachers here, an analysis will be presented using AFMIS expenditure data. AFMIS expenditure data provides an estimate of the relative distribution of haquzama teachers. AFMIS data is considered more reliable and comprehensive but a key issue with the dataset is that a single expenditure code (21124 ‘overtime’) includes expenditures for: (i) overtime payments (by non-teacher MoE staff); (ii) haquzama inside tashkeel salaries; and (iii) haquzama outside tashkeel salaries. As such, the expenditure reflects not only teacher FTEs but also overtime payment of PED and DED staff49. 190. Overtime expenditures as a percentage of teacher salary expenditures vary significantly across provinces. Nationwide the MoE spent 17.9 billion AFN (256 million USD) on teacher salaries and 3.6 billion AFN (52 million USD) on ‘overtime’ (haquzama salaries plus overtime payments of MoE administrative staff) in 2016. Thus nation wide the MoE spent an additional 20% on overtime compared to teacher salaries. Overtime expenditure rates (as percentage of teacher salaries) vary from 4%-7% in Daikundi, Paktika, Uruzgan, and Badakhshan at the lower end, to 48%-59% in Kabul Province, Panjshir, Kandahar and Zabul at the higher end, see Map 5. 49Note here that the number of non-teacher positions eligible for overtime is limited and small compared to the teacher workforce. As such, the amount that can be spend on overtime payments to PED and DED staff is limited, and the bulk of overtime expenditures should reflect haquzama salaries. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 69 Table 10: Overview of different haquzama datasets and AFMIS estimate 2017 Haquzama Outside tashkeel Inside tashkeel Total estimate Payroll Budget Academic Budget Academic AFMIS overtime expenditure Data Source Registration Unit Supervision Unit Supervision estimate50 Database Dataset Dataset Dataset Dataset Kabul Province 1.467 1.075 1.015 1.818 1.608 2.704 Kapisa 550 191 N/A 16 N/A 1.054 Parwan 1 1.528 1.069 119 61 1.648 Wardak 8 21 5 594 1.268 793 Logar 384 389 392 48 45 561 Ghazni 20 126 685 1.470 6.096 1.585 Paktia 0 474 N/A 432 N/A 933 Nangarhar 42 0 N/A 2.520 N/A 2.445 Laghman 154 23 16 45 44 622 Kunar 312 339 N/A 103 N/A 810 Badakhshan 131 43 0 10 801 682 Takhar 726 801 703 555 1.339 1.711 Baghlan 5 350 N/A 291 N/A 1.053 Kunduz 0 501 N/A 14 N/A 684 Samangan 1 5 5 815 1.196 772 Balkh 371 0 0 1.331 1.141 2.324 Jawzjan 367 225 215 590 909 763 Faryab 70 27 N/A 165 N/A 396 Badghis 119 181 N/A 65 N/A 452 Herat 2.207 1.004 N/A 452 N/A 3.164 Farah 2 374 N/A 171 N/A 463 Nimroz 0 111 23 443 557 549 Helmand 480 267 437 263 334 235 Kandahar 1.439 2.228 1.435 1.472 1.258 2.341 Zabul 26 250 252 0 0 184 Uruzgan 1 16 2 24 21 90 Ghor 181 105 1.266 28 516 1.721 Bamyan 230 262 9 1.364 1.483 286 Paktika 25 78 45 594 518 187 Nuristan 28 25 13 0 4 198 Sar i Pul 54 34 23 586 738 815 Khost 1.777 1.790 1.657 811 843 1.875 Panjshir 417 426 N/A 271 N/A 929 Daikundi 0 74 56 1.454 1.657 367 Kabul City 2.095 1.277 N/A 2.270 N/A 4.316 Subtotal 13.690 14.620 9.323 21.204 22.437 - Total 31.760 N/A 39.714 35.824 (Source: Payroll registration Database, Budget Unit, Academic Supervision Directorate and AFMIS 2017) 50Estimate is based on 2017 AFMIS total overtime expenditures, multiplied by 80% (part of MoE workforce working as teachers) and divided by 56,700 AFN (estimate of annual Haquzama salary of average education level). Estimate was halved for Kabul City given the considerable number of additional employees in the capital. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 70 Map 5: Provincial PTRs and overtime expenditure rate (as percentage of teacher salary expenditures), 2016 (PTRs (enrolled/tashkeel in color, and overtime expenditure rate in digits) 23% 21% 11% 7% 18% 13% 24% 16% 29% 53% 11% 10% 44% 30% 26% 16% 13% 48% 15% 15% 13% 13% 36% 4% 21% 20% 40% 22% KABUL CITY 5% 20% NATIONAL AVERAGE 4% 16% 59% PTR 30-34 14% 54% PTR 35-39 21% PTR 40-44 PTR 45-49 PTR 50-54 PTR 55-59 PTR >60 (Source: EMIS, Payroll Registration Database, AFMIS) 191. Overtime expenditures (as an estimate of haquzama teachers) are not prioritized across provinces with high PTRs. Whereas some province with high PTRs had high rates of overtime expenditures (Khost, Zabul, Kabul Province, and Kanadahar), others did not (Helmand, Ghazni, Herat, Badghis, and Daikudi). Conversely some provinces with very low PTRs still had very high rates of overtime expenditures (Panjshir, Parwan, Bamyan, Kapisa). Most notably, Panjshir province spent 118 million AFN (1.7 million USD) on teacher salaries in 2016, and an additional 63 million AFN (0.9 million USD) on overtime, even though Panjshir already had the lowest PTR in the country (based on enrolled students per tashkeel teacher). 192. As a result, provincial discrepancies in NET PTRs (attending students per estimate teacher FTE) are not reduced by the allocation of temporary teacher positions. Some provinces still have twice as many students attending per teacher FTE, as other provinces. In the most extreme ends, Helmand (NET PTR 42) has three times more attending students per teacher FTE, than Panjshir (NET PTR 14), see Map 6. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 71 Map 6: Provincial NET PTRs (number of attending students per teacher FTE, based on AFMIS estimate) 2017 27 29 34 23 27 30 25 23 28 14 20 32 27 21 24 26 28 30 22 33 32 26 24 35 30 34 38 31 KABUL CITY 30 29 NATIONAL AVERAGE 28 27 32 42 29 26 (Source: Payroll Registration Database, EMIS, AFMIS) 193. There is no school level data on the distribution of haquzama teachers across schools. However, for this study, two school principals (of large high schools in Kabul City) were interviewed. Both principals provided a summary of key indicators (number of students per grade, number of tashkeel teachers by gender, and support staff), as well as their number of haquzama contracts. The two schools were similar on key indicators but one school had significantly more haquzama teachers. The principals were unable to explain or even guess what could be the reason for this discrepancy. This seems to suggest that there is no clear policy or regulation on the number of haquzama contracts and how they are prioritized across schools and provinces. 3.4.4 Equalizing Pupil-Teacher Ratios 194. Provincial NET PTRs can be equalized (almost fully) by only redistributing temporary haquzama contracts based on the relative shortage in each province. To balance the number of attending students per teacher FTE across provinces, it is not necessary to terminate or relocate tashkeel teachers. Only seven PEDs (Parwan, Wardak, Badakhshan, Baghlan, Bamiyan, Nuristan and Panjshir) currently have more tashkeel teachers than required based on the national average of attending students (29.8) per teacher FTE. If the estimated 35,824 temporary haquzama positions would be distributed across the remaining 28 PEDs, based on their relative shortages to the 29.8 average, all 28 PEDs would fall into the 30 to 31 NET PTR range, see Table 11. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 72 Table 11: Example of haquzama position distribution based on relative shortages and new NET PTRs Redistribution of Haquzama positions based on relative shortage to national average 2017 Attending Tashkeel Haquzama Total Teacher NET PTR Students Teachers allocation FTE Kabul Province 253.918 5.779 2.418 8.197 31,0 Kapisa 112.122 3.553 185 3.738 30,0 Parwan 166.600 6.217 0 6.217 26,8 Wardak 118.319 4.613 0 4.613 25,6 Logar 100.362 3.260 95 3.355 29,9 Ghazni 282.898 6.726 2.440 9.166 30,9 Paktya 144.153 3.802 913 4.715 30,6 Nangarhar 570.438 14.409 4.173 18.582 30,7 Laghman 143.511 4.506 273 4.779 30,0 Kunar 132.451 4.136 272 4.408 30,0 Badakhshan 260.805 10.503 0 10.503 24,8 Takhar 282.939 8.446 924 9.370 30,2 Baghlan 246.368 9.379 0 9.379 26,3 Kunduz 262.906 6.923 1.675 8.598 30,6 Samangan 87.109 2.610 276 2.886 30,2 Balkh 426.819 12.092 1.967 14.059 30,4 Jawzjan 150.512 4.644 359 5.003 30,1 Faryab 235.725 7.420 432 7.852 30,0 Badghis 97.979 2.567 636 3.203 30,6 Herat 611.907 15.845 4.134 19.979 30,6 Farah 88.127 2.746 186 2.932 30,1 Nimroz 52.019 1.436 273 1.709 30,4 Helmand 126.918 2.780 1.304 4.084 31,1 Kandahar 193.506 4.156 2.061 6.217 31,1 Zabul 38.568 1.024 238 1.262 30,6 Uruzgan 50.772 1.604 88 1.692 30,0 Ghor 149.856 4.437 522 4.959 30,2 Bamiyan 102.427 3.496 0 3.496 29,3 Paktika 102.978 3.395 53 3.448 29,9 Nuristan 34.442 1.484 0 1.484 23,2 Sar-e-Pul 119.045 3.396 528 3.924 30,3 Khost 241.856 4.441 3.241 7.682 31,5 Panjshir 31.850 1.347 0 1.347 23,6 Daikundi 125.322 3.174 909 4.083 30,7 Kabul City 990.798 27.295 5.249 32.544 30,4 Total 7.136.325 203.641 35.824 239.465 29,8 (Source: Staff calculations) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 73 195. The remaining oversupply of tashkeel teachers (in the aforementioned seven PEDs) could be addressed by not allocating any new tashkeel positions in the next 5 years. Provincial NET PTRs can be fully equalized after 5 years, by prioritizing new tashkeel position only to those provinces with current shortages. Assuming 3.9% student and teacher growth, those provinces with a current oversupply of tashkeel teachers (of around 20% more than FTEs needed) will meet the demand for these FTEs in the next five years. 3.5 Discussion and recommendations 196. Current methods used to assess teacher needs, and distribute them across provinces and schools, are not standardized, and not institutionalized. Existing norms on teacher needs are ambiguous, lack detail, and are not consistently applied. As a result, schools and provinces with very similar compositions can calculate considerably different numbers of teachers needed. The lack of clear criteria for the creation of classes, and the apparent variation of applied norms for creating classes is particularly problematic, as the number of classes is the primary driver used to calculate teacher needs. 197. MoE teacher needs calculations are not transparent, and not replicable. This thwarts both a political and technical debate about the necessity of more teachers. More teachers are in principle always better, and in that sense, there is always a need for more teachers. However, a political dialogue is required to determine to appropriate balance between teacher needs and fiscal constraints. Likewise, the technical evaluation of teacher needs can promote fairness and efficiency. However, the current lack of clear norms, transparency, and standardization of methods, prevents education stakeholders from evaluating the necessity of more teachers. 198. Current methods result in significant discrepancies between provinces and extreme discrepancies between schools (or at the least do not reduce these discrepancies). Some provinces have twice the number of students per teacher than other provinces, and some schools have a PTR ten times that of schools with small classes. This is neither fair nor efficient. Afghan children sharing a teacher with a hundred other students do net get the same learning opportunities as children having to share their teacher with only five or ten other students. Furthermore, empirical evidence has shown the significance of PTRs on student and school performance51. 199. This suggests that the MoE can significantly improve both fairness and efficiency in the public education sector, by adopting norms and methods that result in a more equitable teacher distribution. Top-down planning using PTR norms for provinces, district and schools, and clear guidelines and policy choices on how to establish classes, can decrease the extreme disparities currently existing. Rather than evaluating the individual needs of over 200,000 classes (‘actual class size’ method), the MoE could significantly increase equity and efficiency if it would develop a much simpler norm on PTR targets per province and district. 200. In addition, the MoE needs to critically evaluate the costs of its minimum benchmark on creating classes. While the aim of this policy (promoting access to education in rural areas) is generally laudable, the costs are considerable. Low minimum benchmarks for creating classes, combined with a relatively high number of small schools, result in an extreme ‘student compensation burden’ for larger schools. The effective minimum benchmark for creating a 51The Impact of Number of Students per Teacher on Student Achievement (2015). N. Cok & C. Belik. Prcedia: Social And Behavioral Sciences 177 (2015) p65-70. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 74 ‘grade 12’ class is 5 attending students per teacher (minimum 10 enrolled students per class, 1.5 teachers per class and 75% student attendance). That is well below the average of 13 students per teacher in high-income countries. The MoE needs to consider the benefits of this policy against the costs of having enormous classes (100+). The issue of enormous classes can be addressed by: (I) raising the minimum benchmarks for creating classes; (II) allowing multi-grade classes (e.g. a single class and teacher for students from grades 1 and 2); (III) absorbing very small classes into different programmatic modalities (e.g. community-based education), or (IV) significantly increasing the number of teachers employed by the MoE. 201. Last, the MoE’s existing flexibility in human resource allocation (i.e. the temporary haquzama teacher contracts) is not optimally used to address Afghanistan’s volatility in education needs. Sustained insecurity, schools closure, sizeable numbers of internally displaced people (IDPs) and migration have all resulted in significant fluctuations in the numbers of students (and corresponding teacher needs) across schools and provinces. The MoE needs a flexible (temporary) contract modality to be able to effectively respond to these fluctuations. However, this study found no evidence of a standardized method, or national policy to allocate temporary teachers to areas where they are most needed. Instead, processes to identify haquzama needs appeared decentralized, ad hoc, and unregulated. Furthermore, the lack of centralized planning, reporting and oversight creates further risks of abuse. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 75 Recommendations – Teacher Needs and Allocation The Ministry of Education should: SHORT TERM: 202. Revise existing norms on teacher (and school staffing) needs per school. Provide clear and detailed instructions for schools, DEDs and PEDs to calculate the number of teachers needed, and identify the order in which type of teachers (subject specialization) should be prioritized, using a step-by-step model with progressively increasing student numbers. Criteria for creating classes should be clarified, as well as minimum and maximum benchmarks for class size. Norms for additional school staff (both academic and non-academic) should also be clarified in a similar step-by-step model. Furthermore, teacher (subject) qualification requirements per grade and school type should be specified. Revised norms should be clearly communicated to all levels of the public education system. 203. Apply a transparent and simplified formula for allocating teachers per province, using number of attending students per grade as the primary driver, and create an adjustment index for the percentage of rural schools (if needed). Develop instructions for PEDs and DEDs to prioritize intra-province distributions of new tashkeel positions, using relative teacher deficiencies as a driver. Publish the formula and methods used to allocate teachers across and within provinces and publish their outcomes, including PTRs, and where temporary teacher are considered. 204. Develop a policy and clear regulations for the contracting of haquzama teachers. Identify the purpose of this modality, and identify financial regulations and reporting requirements. Centralize planning and allocation of haquzama teachers, based on teacher needs assessments and sudden fluctuations in teacher needs. Utilize the flexibility of these contracts to address short term fluctuations, as well as work towards medium-term equity of the teacher distribution across provinces and schools. 205. Address PTR discrepancies between provinces and between schools, using the revised norms to identify excess and deficiencies. In the short term, relocate tashkeel teachers wherever possible, and use temporary teacher contracts to address severe deficiencies (i.e. focus on enormous classes). 206. Revise the existing policy toward very small classes. Increase the minimum benchmark, allow for multi-grade classes in small rural schools, or integrate small rural schools into a different programmatic modality. Furthermore, enforce regulation, so that very small classes are only created in rural areas to promote access to education, and not in semi-rural and urban areas. MEDIUM TERM: 207. Progressively address PTR discrepancies between provinces and between schools. In the medium term, using attrition, recruitment and the allocation of new tashkeel position to address discrepancies. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 76 4 Teacher Recruitment 208. Effective and merit-based teacher recruitment is a critical tool for promoting teacher quality. Given the semi-permanent nature of civil servant appointments and low levels of attrition, it is key to recruit the most qualified candidates into the teacher workforce. In addition to promoting overall teacher quality through merit-based selection (and addressing corruption, nepotism and patronage), an effective teacher recruitment policy can also be applied to address gender or geographic disparities in the teacher pool. For example, a recruitment policy can aim to promote female teacher recruitment or promote recruitment opportunities for qualified candidates in remote areas. 209. In the past years however, the merit-based recruitment of new teachers appears to have been significantly undermined by nepotism, corruption and unlawful interference, resulting in the Afghan people’s diminishing trust in the education system. A 2017 corruption vulnerability assessment report of the Independent Joint Anti-Corruption Monitoring and Evaluation Committee (MEC) singled-out teacher recruitment as the most damaging area of corruption: “One corruption vulnerability emerged as being the most serious – the widespread, countrywide appointment of teachers on the basis of influence, or nepotism and bribery, not on the basis of merit. In short, teacher appointment is largely corrupted. This is the most damaging issue for the education of students in the country.�52 210. This chapter will discuss teacher recruitment as it took place in the past years, the inherent limitations of this, and the critique it has received in recent years. Thereafter, it will describe the proposed reforms in teacher (and civil service) recruitment (implemented in late 2017, and early 2018), and describe the new teacher recruitment step-by-step. The study aimed to look critically at the recent reforms in teacher recruitment, in order to identify areas for further improvement. However, this study was unable to fully and accurately evaluate the effectiveness of these reforms, due to ongoing implementation of these reforms and inherent limitations of this study. This chapter ends with a discussion on these reforms and provides recommendations for further improvement. A full and complete evaluation of the new recruitment process would enable a more accurate assessment of how effective these reforms have been. 4.1 The Old Teacher Recruitment Process 211. In the past years, teacher recruitment was managed and implemented at the provincial level. Candidates applied for teacher positions and took teacher entry exams in the provincial capitals. The process was managed and implement by the provincial education directorates (PEDs), and monitored by provincial authorities including the provincial directorates of the Independent Administrative Reform and Civil Service Commission (IARCSC). The role of Kabul, and the Ministry of Education (in Kabul) was mostly limited to preparing the (open-ended) teacher entry exam. The MoE General Directorate of Academic Supervision prepared the entry exam materials, and MoE delegations travelled to the provinces to deliver the testing material and monitor the entry exam. Scoring of the exams mostly took place in the provinces, although in some cases central MoE staff was also involved in exam scoring53. 52 Ministry-wide Vulnerability to Corruption Assessment of the Ministry of Education , Independent Joint Anti- Corruption Monitoring and Evaluation Committee (October 2017), p2. 53 Education Sub-National Assessment: Strengthening Public Education Management . p.62. Altai Consulting. (2015). Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 77 4.1.1 Critique of corruption 212. PEDs openly deviated from national civil service regulations54, and the recruitment process appeared to be vulnerable to corruption at almost every step. National regulations regarding civil servant (and thus teacher) recruitment are ‘fairly robust’. However, the decentralized implementation of the teacher recruitment process and limited oversight from Kabul resulted in significant deviations from these national standards. “The PEDs did everything themselves, shortlisting, scoring, selection and appointments – there was no oversight from the Ministry and nobody was checking what they were doing.�55 213. Many MoE officials at national, provincial and school level, interviewed for this study, expressed their believes that abuse of teacher recruitment was widespread. Teachers interviewed, confided they heard of colleagues making payments in order to be appointed. Officials from the central Ministry in Kabul stressed their disdain from recruitment practices in the provinces: “Teacher recruitment was delegated to the PEDs, and most of them misused their authorities resulting in inefficiency. PEDs were influenced by MPs and local powerbrokers, in particular during teacher recruitment. Nepotism was common.�56 214. Already in 2015, a MEC report stressed the vulnerabilities to corruption across all steps of the teacher recruitment process. From cheating on the entry exam, the distribution of the questions or answers before examination, the forging of exam scores and application documents, to the last minute adding of names to the list of successful applicants57. Similarly, a 2015 assessment of provincial education directorates, reported significant variations in teacher recruitment processes across PEDs, inconsistent implementation of regulations, perceived corruption, and the need to rigorously audit provincially managed teacher recruitment processes58. 4.1.2 Lack of data on recruitment 215. In addition to the decentralized implementation of recruitment, and the lack of oversight, there was no centralized systematic information flow on the quality or quantity of teacher recruitment. As a result, there is no quantitative data available in Kabul on new teacher recruits (i.e. the number of new recruits and their qualifications). The research team was not able to find any information on the number of teachers hired in the past years through the recruitment process, nor any details (e.g. gender, age, qualification) on the newly recruited teachers. Knowledge of teacher recruitment at the MoE in Kabul is limited to anecdotal references (e.g. “there was a lot of abuse� or “that province was not able to fill all its vacancies�). 54 Ibid.p59. Most PEDs did not archive all application documents and entry exams, as per the civil service regulations. Provinces also varied in terms of who graded the exams, who was on the teacher recruitment committee, how results were announced, and how successful candidates were selected. PEDs openly disregarded Ministry directives regarding teacher recruitment, see: “Vulnerability to Corruption Assessment of Teacher Recruitment in the Ministry of Education�, Independent Joint Anti-Corruption Monitoring and Evaluation Committee (June 2015), p5. 55 Kabul based senior MoE official interviewed for this study. 56 Kabul based senior MoE official interviewed for this study. 57 Vulnerability to Corruption Assessment of Teacher Recruitment in the Ministry of Education. Independent Joint Anti- Corruption Monitoring and Evaluation Committee (June 2015), p7-12. 58 Altai Consulting. (2015). Education Sub-National Assessment: Strengthening Public Education Management . p.63. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 78 216. Furthermore, there is no information available on the quality of the recruitment process (fairness, transparency, merit-based, etc.). The MoE (GDHR) does not centrally collect recruitment documentation (e.g. application documents, teacher entry exams) nor is there any evidence that the GDHR systematically evaluated or audited the teacher recruitment process in the provinces. Besides the central Ministry representatives providing the exam and overseeing the examination, the role of the Ministry has been minimal59. There seems to have been no central-level evaluation of the quality of the recruitment process in the provinces, or if the PEDs adhered to recruitment regulations. 217. Without any information on recruitment, the MoE in Kabul also does not analyze the teacher recruitment process, such as the number of applicants, or the exam scores per province. Analyzing exam scores, and the quality and quantity of applicants provides insights in regional, rural and gender disparities of teacher applicants and the overall attractiveness of the remuneration package. In fact, analysis of recruitment data is essential for developing an effective teacher policy that should inform teacher training, teacher allocation, and teacher remuneration. 4.2 The New Teacher Recruitment Process 218. On June 20th, 2017, H.E. President Ghani announced significant reforms in the teacher recruitment process60. Henceforth, applicants for teacher positions will have to complete a collective competitive entry exam, similar to the kankor university entry exam61, and future teacher recruitment will be managed by the IARCSC with support from the MoHE. This new recruitment process is regulated by a ‘Bill of Collective Examination’ 62, in addition to existing civil servant rules and regulations. This ‘new’ teacher recruitment process differs in significant ways from past recruitment practices. 219. First, the IARCSC is the lead implementing agency for the new teacher recruitment, with a significant role for the MoHE to support this. The role of the MoE and PEDs is limited to identifying vacancies, developing ToRs and supporting/facilitating teacher recruitment. Whereas in the past, the role of the IARCSC was to oversee the teacher recruitment process at the PEDs (as the procedural and regulatory guardian), the commission (with support from MoHE) is now leading the recruitment process from the initial announcement all the way to selecting the successful candidates. The overall recruitment process is managed by two committees within the IARCSC, a steering committee and an implementation committee63. 59 Some PEDs reported that the central line Ministry is involved in the scoring of the exam, ESNA REF. 60 Presidential Decree #1160, serves: to bring reform and improvement in the civil servant recruitment process (including teachers) in ministries and other government organizations by allowing opportunities for talented youth, and by adopting the collective competitive examination system, administered by the Independent Administrative Reform and Civil Service Commission. 61 ‘Kankor’ is derived from the French ‘concours’ which translates to ‘competition’. Kankor exam translates to ‘competitive exam’. The kankor exam in Afghanistan refers to the university entry exam, where examinees with the highest exam scores could gain (i.e. win through competition) access to the state’s universities. Historically the kankor exam is considered to uphold the highest standards of quality and impartiality. The kankor exam was the first to introduce multiple choice response formats and automated scoring. The collective competitive exam method applies not only to new teachers, but also to the recruitment of other ‘new’ civil servants, as well as the filling of existing vacancies and the replacement of acting civil servants. 62 In addition to the existing Civil Service Law (2005) and civil servants law (2008). 63 Steering Committee Members: Deputy Minister of administration and finance of all government ministries Heads of HR Departments of all ministries and independent organizations Representatives of Civil Society Organizations Implementation Committee Members: MOE (the beneficiary) IARCSC (the lead and monitoring organization for Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 79 220. Second, almost all parts of the recruitment process are now centralized in Kabul . In the past, announcements, collection of applications, and shortlisting was executed by the PEDs, and most PEDs also scored the exams and selected successful candidates. The role of the MoE (in Kabul) was limited to preparing the exams (and in some cases scoring and selection). In the new recruitment process, provincial entities primarily play a supporting role, and only those parts that cannot be centralized (filling of application documents, taking the entry exam, and collecting biometric data) take place in the provincial capitals. 221. And third, the teacher entry examination process is implemented in the same way, and governed by the same rules, as the kankor university entry exam. That means that examinees will receive electronically generated randomized questionnaires, with multiple-choice response formats and exams will be electronically scored. Applicants sitting the exam will receive a unique examinee ID code that links their identity to their questionnaire and response sheet. This will significantly reduce the opportunity to cheat during the exam, leak exam answers prior to the exam, or manipulate exam scores afterwards. 222. Early 2018, the IARCSC, and the MoHE were in the process of hiring 17,700 additional civil servants through this new ‘competitive’ recruitment process. This number includes 8,000 new teachers and almost 4,990 additional MoE staff64. At the time of writing (February/March 2018), the IARCSC had just finalized the application and shortlisting process and the first teacher entry exams took place in the East and South of the country. It is not clear exactly when the first round of this ‘new’ teacher recruitment will be finalized. However, this should be finalized in the first half of 2018. The below section describes the new teacher recruitment process in more detail, and discusses its shortcomings and potential for improvement. 4.3 The New Teacher Recruitment Step-By-Step 223. The new recruitment process is described in the following nine steps: (I) Vacancy Identification; (II) Announcement; (III) Application; (IV) Shortlisting; (V) Exam ID and biometric data collection; (VI) Examination; (VII) Scoring; (VIII) Preparing list of successful candidates; and (IX) Review of qualification documents and selection. I. Vacancy identification 224. Vacancies are identified by the line ministries. In the case of teachers, vacancies were identified by the MoE. According to MoE officials the list of vacancies was prepared through an internal ‘bottom up’ consultation, using data submitted by the schools. The results is a specified list of vacancies by type (e.g. physics, Dari, Pashto teacher), number, per province, district and school (but not by gender). The MoE GDHR (central level) prepared two ToRs for the vacancies. One for under qualified teachers (below ‘grade 14’ educated) and one for ‘grade 14 education teacher positions. A copy of the ToRs can be found in Annex 2. The MoE submitted the list of vacancies and the two teacher ToRs to the IARCSC (central level). recruitment), IDLG (supporting the recruitment process in provinces: technical and space provider for recruitment examination) Women affairs organizations networks (to assess and monitor the recruitment process from gender perspectives) Donors representatives (to oversee transparency and effectiveness in the recruitment process) MOHE/Kankor (technical support: digital examination) MOHE/universities teachers (preparation of questions for the examinations) Science Academy (review and preparation of questions) Representatives of each ministry (to oversee the recruitment process). 64 The 4,990 ‘additional MoE staff’ includes managerial and administration staff, support staff, but also new teacher positions. The IARCSC was unable to specify how many of the 4,990 positions were teacher positions. According to the IARCSC all 17,700 positions are bast 5 and 6. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 80 II. Announcement 225. The new round of recruitment was announced trough TV, Radio, the AIRCSC website and hardcopy posters at the IARCSC offices. Announcements were also made through the provincial and district offices of governors, and the line Ministries’ directorates. Non- government institutions (e.g. Mosques) were also involved in announcing the upcoming process. As key dates of the recruitment process were not known, these were not included in the announcement. Rather it announced an upcoming round of civil servant (and teacher) recruitment, and specified the number of vacancies in each province (but did not specify gender, the IARCSC stressed this would be against the civil service law). The IARCSC prepared a template for the announcement of vacant positions that was used by each province to indicate for which positions candidates could apply. The announcement did not describe how recruitment procedures had changed since the last round. III. Application 226. The IARCSC in Kabul managed the application document distribution and collection process. The IARCSC started the distribution of application forms in October 2017. It distributed more than 300,000 application forms (there is only a single version of the application form, and copies can be made in case more forms are required). The application forms were made available at provincial- and district governor’s offices, provincial- and district line directorates as well as offices of the IARCSC. According to the IARCSC all 400 districts received copies, despite security challenges. In addition to the hard-copy forms, teacher candidates could apply online through a web-portal. 227. On the form (see Annex 3), the applicant indicates the province and district where he or she wants to apply. The form does not indicate the specific vacancy applied for (i.e. the position and school). The applicant can only apply (for vacancies) in the province where the application is made65. Furthermore, the applicant indicates his or her education level and subject specification (e.g. TTC trained chemistry teacher). The applicant also attaches a passport size photo to the application form. Importantly, the application form did not include instructions on the filling of the form, nor an explanation or description of the recruitment process. 228. The IARCSC received approximately 270,000 applications for teacher and non-teacher vacancies. This included around 30,000 online applications. The IARCSC sorted the application documents according to three main criteria: (i) language (candidates can choose to take the exam in Pashto or Dari); (ii) field of study (e.g. TTC science graduate); and (iii) type of position applied for (e.g. chemistry teacher), specified by location. Furthermore, MoE General Education applications were organized separately form TVET (and other program) applications. Online applications were printed by the IARCSC and physically organized with the hard copy applications. 65This was said to prevent inter-provincial competition. That is, to prevent the better educated urban province applicants from competing with lesser-qualified rural province candidates, and to maintain a fair provincial balance in the employment created by the vacancies. However, since the applicant’s home address is not registered on the application form (or anywhere else), candidates can still apply for any position in the country. Applicants do have to apply (i.e. submit their form) in the province where they want to work and sit the exam there. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 81 IV. Shortlisting 229. Between late 2017 and January 2018, while tranches of hard-copy applications were arriving at the IARCSC office in Kabul, the commission was filtering the documents and disqualifying candidates that were not eligible (i.e. did not meet the minimum education criteria or errors in filling out the application forms, etc.). Furthermore, the commission found a large number of candidates that applied multiple times (for different positions and/or different provinces) and removed ‘duplicate’ applicants from the shortlist (applicants were allowed to apply on one vacancy only). On the 24th of January 2018, the commission published the names of all applicants approved and shortlisted for the teacher entry exam (as well as the exams for other civil servant positions) on its website. 230. This applicant ‘shortlist’ included 118,544 applicants for teacher positions that were organized in data files by province (an additional 106,778 applicants were shortlisted for the remaining 9,700 civil servant positions). The IARCSC also published a ‘hard-copy’ shortlist at the offices of the provincial and district governors. According to the commission, some 19,000 applicants were not eligible to apply and some twenty to thirty thousand duplicates were found and removed from the pool. For a period of five days after publication, rejected candidates were allowed to submit their complaints and challenge the decision of the commission. 231. While the IARCSC was organizing and filtering applications, the MoHE directorate of examination prepared a Student Data Entry Database (SDED). This database registers the name of each of the examinees, as well as their qualifications (i.e. training or education) and the field and positions they are applying for. Every batch of application documents was verified in the shortlisting step and then entered into this database. Once an examinee’s data is entered into the SDED, the database generates a unique ID number for each examinee (Tazkeera ID numbers are not used in this process). This exam ID number is printed on the exam ID cards, questionnaires, answer sheets, and linked to the collection of biometric data. V. Exam ID and Biometric data collection 232. Biometric data was collected when the examinee picked up his/her exam ID card. Province by province, shortlisted candidates were called upon to collect their exam ID card (with the unique ID number and photo for each examinee). Exam ID cards were collected from the provincial capitals only. According to one IARCSC official, candidates received instructions on the examination and recruitment process on the day of examination. When receiving their exam ID cards, examinees also had to submit their biometric data. Biometric data and the unique ID number were thereafter linked. Biometric data collection included fingerprints and an iris (eye) scan. The IARCSC indicated that the biometric data collection was intended to verify the examinees identity during examination and might be used later for other HR management purposes. 233. Not all shortlisted candidates collected their exam ID card and were recorded in the biometric data system. For example in Kandahar, where 2,806 candidates were shortlisted (for both teacher and non-teacher vacancies), only 1,657 candidates (59%) were registered in the biometric data system. While biometric data collection and entry examination was ongoing at the time of writing this report, comprehensive nation-wide data on registration and applicant turn-out was not yet available. However, initial information released by the IARCSC on teacher entry examination in four southern provinces (where exams took place in February 2018), Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 82 suggested that between 50% and 65% of shortlisted candidates collect their exam ID card66. It is not clear why a significant percentage of shortlisted candidates (41% in Kandahar) were not registered in the biometric data system. VI. Examination 234. The teacher entry exams will be organized in all 34 provincial capitals. Nation-wide the IARCSC aimed to organize the exams in an as short as possible time frame. The first teacher (and civil servant) entry exams were organized in the East and South of the country. The commission aimed to organize all the different exams (e.g. for all teacher categories and vacancies at line ministries) in a single day. However, in Kabul, given the high number of applicants (more than 55,000), it expects to require three days to complete all exams. Wherever possible, the entry exam will take place in classrooms of public universities (facilitated by the MoHE), in school classrooms, or outside. Examinees will be placed in a specific order, and observers will assure the integrity of the exams. 235. At the exam location, examinees will be required to show their exam ID and again submit their biometric data for verification. According to the IARCSC, people without an exam ID card will not be allowed access to the exam location on the day of the exam, to reduce opportunities of interfering in the examination process: “Even the provincial governor will not be allowed to enter the exam location on the day of the teacher entry exam�.67 236. Examinees will also sign an attendance sheet. This attendance sheet includes their name, father’s name, exam ID number, and the photograph of the candidate, and this will be checked against the candidate’s exam ID card and his/her actual appearance. When receiving the exam questionnaire and answer sheet, the examinee will also receive a description of the examination and recruitment process. According to the IARCSC, the examinee will also be required to sign a ‘commitment letter’ stating that he/she is aware of the examination and recruitment process, as well as the minimum eligibility criteria. 237. Furthermore, candidates need to indicate the specific position applied for on their answer sheet. Thus all answer sheets include a numerically coded list of all districts, schools, and subjects for which there are vacancies in a given province. Examinees identify their choice through the marking of the corresponding numerical code. Candidates should apply for one position only. 238. The MoHE directorate of examination has prepared a ‘Question Bank Database’ (QBD) that includes all questions for all exams. The QBD includes the questions for 64 different types of civil servant entry exams. This database is not connected to the examinee database (SDED). However, once information is transferred from the SDED (with examinee information), the QBD can generate a unique questionnaire for each examinee, related to his or her subject of application (e.g. high school chemistry teacher). 239. Questions for the QBD have been prepared by an ‘Academic Committee’. This committee consists of 30 professors from four universities in Kabul (Kabul University, Kabul Polytechnic, Kabul Medical, and Kabul Education), as well as Ghazni-, and Nanagarhar University. The academic committee members were allowed to request external support materials (e.g. 66 “IARCSC holds examinations in Zabul, Kandahar, Uruzgan And Nimruz �, IARCSC press release posted on their website February 17, 2018. 67 Senior IARCSC official interviewed for this study Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 83 documents, references, old exam questionnaires, expertise, etc.) to develop the questions. But nobody outside of the academic committee was allowed to draft questions. In previous years, the MoE’s academic supervisors and lecturers of the TTCs were involved in the drafting of teacher exam questions, but this is no longer the case in the new process. The question bank was finalized late December 2017. 240. According to the IARCSC, each examinee will be given a unique exam questionnaire and answer sheet. Each questionnaire will include between 100 and 160 multiple-choice questions (depending on the type of exam). Most questions are subject area specific and depend on the position the examinee is applying for, whereas around 25% of the questions are generic (i.e. testing general knowledge of language or math). Within an exam category (e.g. ‘entry exam for high school chemistry teacher’), question and answer formats are jumbled (i.e. presented in a randomized order). In addition, applicants sitting the exam for the same position will not receive the exact same set of questions. Instead, the examinee will receive a random set of questions (within his/her type of exam) out of the question bank. According to the IARCSC, the system generates the same proportion of easy and difficult questions for each questionnaire. Supposedly equal but not the same. One IARCSC official indicated, that the commission was legally prohibited from sharing any sample exam questionnaire or individual question from the QBD for the purpose of this study (or any other purpose for that matter). And therefore, the quality of the exam questions could not be assessed. VII. Scoring 241. All answer sheets have been (or will be) returned to Kabul, where they will be automatically scored by MoHE using ‘Optical Mark Readers’ (OMRs, the same machines used by MoHE to score the kankor exam). However, before scoring, all collected answer sheets get scanned and electronically saved in the MoHE (directorate of examination) database, to keep a record of the original answers of each candidate. If candidates disagree with their results or ask for the rechecking of their answer results, the MoHE will cross check the score of a candidate with his/her original answer sheet. The OMR machines can read and score up to eleven thousand answers per hour. Given that each questionnaire is unique (jumbled order and not the same questions), there is no answer key for any type of exam. The IARCSC also does not publish the questions or the answers on their website, and thus it is not possible for examinees to independently verify their own exam score. VIII. Preparing list of successful candidates 242. The list of successful candidates is prepared by MoHE, based on the highest exam scores within a single vacancy or pool of vacancies. A candidate thus competes for a specific position at a specific school against all other candidates that applied for that same specific position and school. This selection method could be labeled ‘single non-transferable application’, as opposed to a ‘pooled’ or ‘cascading’ selection method, where the highest scoring candidates are given first choice out of a large number (pool) of possible positions. The chosen selection method has its strengths and weaknesses, which will be discussed in the following section. 243. Exam scores and outcomes of the teacher entry exam will be made available through a web application (on the IARCSC website). Examinees can consult their exam score and outcome by entering their exam ID number in the application. The application will then produce a personal overview (name, father’s name, province, and position applied for) as well as their exam score, and an indication if they have been selected or rejected. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 84 IX. Review of qualification documents and selection 244. Based on the MoHE lists of highest scoring candidates, the IARCSC will invite successful candidates to submit their education certificates for verification. For example, for a given school in Herat Province, there are two vacant positions for chemistry teachers, and there are about 40 candidates who competed for them (by marking this school’s code in Herat and marking ‘chemistry teacher’). The top two candidates will be offered the two positions. MoHE will send the list of the top two candidates (plus reserves) to AIRCSC, and the IARCSC invites them to submit their certificates for verification. If the top two candidates do not have the education certificates they claimed to have on their application form, or if they do not want the position, they will be ‘disqualified’ and another two reserved candidates will be invited by the IARCSC. The process will be repeated until candidates are selected for all positions. In practice, the reserve candidates are shifted or sent to those posts that are not accepted by successful candidates. The selected teacher candidates will be introduced by IARCSC to the MoE’s GDHR. 4.4 Discussion and Recommendations 245. This section discusses the recent reforms in the teacher recruitment process, and the (ongoing) recruitment of new teachers in the first half of 2018. With the new teacher recruitment process ongoing at the time of writing this report, it is important to stress that this study was unable to fully evaluate the impact of the reforms, due to ongoing implementation and the inherent limitations of this study. However, this section provides an initial assessment of the progress made on addressing corruption vulnerabilities in the recruitment process, the logistical challenges of implementing large scale, and nation-wide recruitment of civil servants, and identifies some of the gaps and weaknesses in the new process. It ends with a set of recommendations. 4.4.1 Key Strengths 246. The new recruitment process addresses some of the most critical corruption vulnerabilities identified in the old process, and it creates opportunities to further improve the recruitment policy and identify regional and gender discrepancies. 247. First, the creation of a centralized electronic database on applicants that registers all steps of an individual in the recruitment process (application, examination, selection) is a tremendous step forward in addressing previous corruption vulnerabilities . The central database allows for auditing of the process, to identify irregularities and significantly reduces (if not completely eliminates) the opportunity for subnational officials to manipulate the outcome of the process. The post-hoc forging of application documents and the last minute adding of names on the list of successful candidates should be easily identifiable in this system and the overall number of people able to interfere in the outcome has been significantly reduced. 248. Second, the new entry examination method almost completely eliminates opportunities of abuse. The randomized questionnaires and answer formats, drawn from a larger question bank reduces the opportunities of cheating during the exam (i.e. copying other examinees answers) and the beforehand sharing of questions or answer keys. Combined with the higher number of multiple-choice questions, and the more rigorous verification of identity (through the exam ID and biometric data collection, reducing the opportunity for impersonating a candidate), the new exam method makes it almost impossible for an unqualified candidate to score high on the teacher entry exam and be selected. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 85 249. Third, the unique and randomized exam questionnaire and answer sheets allow for a gradual (province by province) and flexible roll out of a nation-wide civil servant recruitment effort. If there would be only one entry exam questionnaire for high school chemistry teachers (or multiple versions of the same exam questionnaire) there is a risk of this questionnaire being shared beforehand with examinees. To counter this, all chemistry teacher applicants would have to take the exam on the same day, and at the same time. This is nearly impossible, given the large number of applicants for teacher positions across the country, the tremendous challenges posed by sustained insecurity in many part of the country, as well as severe climate conditions restricting mobility in rural areas (particularly in the winter season). The randomized questionnaires (if drawn from a large enough question bank) allow for the flexible and gradual implementation of teacher recruitment (i.e. province by province or region by region). 250. Fourth, the ‘single non-transferable application’ selection method (a group of applicants only compete for a single vacancy) eliminates opportunities for favoritism in the final selection and appointment stage. In the old recruitment process, the final stage of selection and appointment was a complex process. Teacher applicants are typically immobile (not able and/or willing to work outside a specific village, city or district), yet they competed for a large pool of vacancies across the province. The resulting ambiguity of combining exams scores with applicant preferences resulted in vulnerabilities for favoritism and corruption. 251. Last, the central electronic database on teacher applicants, combined with exam scores, provides tremendous information to further improve teacher policies. The database is perhaps the single largest Afghan dataset on human resources availability across regions, genders, and subject specialization areas68. Thorough analyses of the applicant database can help inform the need for teacher training facilities (TTCs, pre- and in-service training) as well as ‘affirmative action’ type policies to promote female recruitment and the recruitment of rural teachers. The database provides information on the number of applicants (willing to work as teachers), by gender, region, provinces, district, village, and by qualification. It could be used to develop flexible remuneration tools, as well as identify areas where tailored policies are required to promote women into the teacher workforce. 4.4.2 Key Weaknesses 252. The implementation of the new teacher recruitment process, led by the IARCSC, has thus far critically lacked transparency and has created new complications that could be considered unfair, inefficient or ineffective. Some of these complications are directly linked to choices for methods in examination and selection, whereas others could be addressed without impeding the strengths of the new process. 253. First, the new teacher recruitment process has lacked transparency and awareness of most stakeholders is limited or absent. Most officials interviewed for this study, including central MoE officials, PED officials, school principals and teachers were unable to describe how the new recruitment process worked. The study was also unable to find an implementation plan or any written document that described the new recruitment process. Most MoE officials were unable to describe what role the Ministry of Education had in this process and stressed that the IARCSC was leading the process with support from MoHE. While the IARCSC has published press releases on exams taking place across the country (February 2018) and published the shortlist of 68The Kankor exam results also provide a large data set, however the IARCSC dataset indicates availability of specific knowledge required for teacher (and other civil servant) positions of those willing to work in these positions. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 86 applicants on its website (January 2018), it has not described how the new recruitment process would be implemented, or what has changed since the last round of teacher recruitment. 254. The IARCSC has not provided adequate instructions to applicants on how the application, exam and selection processes work. The application document lacked an adequate instruction on the criteria and rules for applying. The IARCSC received more then 20,000 applications of candidates trying to apply for multiple positions, apparently unaware that this was not allowed. In interviews, IARCSC officials were ambiguous about when applicants would be given instruction and explanation about the recruitment process. Instructions would be given during the exam, possibly when exam IDs would be distributed and potentially also via the IARCSC ‘Facebook’ page. 255. Instruction is particularly relevant in the light of the ‘single non -transferable application’ selection method. Given that an applicant can only apply to a single position, and competes with all other applicants for that position, the choice for a specific vacancy significantly alters the chance of being successfully selected. Applying for a ‘popular’ position (e.g. a Dari teacher position at a reputable school in a developed, safe urban area) might mean that the applicant is competing with dozens of other candidates, significantly reducing his or her chances. Whereas if the applicant applies for an unpopular position, there might be little or no competition at all. This however, is not explained to candidates, until during the exam when the applicant also marks his or her specific choice for school and vacancy. 256. From the perspective of the applicant it is difficult to verify the exam score or the fairness of the selection process. Applicants can find their exam score on an IARCSC web application, however they have only access to their own score. As such, they do not know if their score is relatively high or low. They also do not have an answer key against which they can assess their score. Therefore they do not know if they lost points due to incorrect answers or unintelligible marking of the answer sheet. An applicant also might find himself rejected for a teacher position while having a relatively high score on the exam (e.g. 85% correct), and then later hear from a different candidate, with a much lower score, being successfully hired in a different position. It is therefore important to transparently explain the selection process, for otherwise this applicant will still perceive the recruitment process as corrupt or unfair. 257. While the new recruitment process seems to have significantly reduced vulnerabilities to actual corruption, it has done little to address the wide spread perception of corruption. The new recruitment process applies robust measures to avoid interference, abuse and favoritism in teacher recruitment, but thus far the IARCSC has done little to explain these measures to the public or potential candidates for teacher positions. Teachers interviewed for this study were all aware of a new recruitment round taking place for a large number of new positions, but none of them was aware that the recruitment process was reformed or that anti- corruption measures would be implemented. Perceived corruption is relevant for it discourages qualified individuals from applying, as they perceive the selection process to be unfair. It is also important for the reputation of the teacher profession in general, and for the critical element of trust between the government and the Afghan people. 258. The new teacher recruitment process lacks a clear description, regulation and accountability mechanism. While teacher recruitment is regulated by the civil service law (and procedures), and the exam is regulated by the ‘Bill of Collective Examination’, some details regarding the new process need to be clarified. For example, what happens to vacancies where nobody applies for? Whom will be given the first opportunity to accept these positions? Also, if entry exams of the same category do not have the same questions, but are ‘equally difficult’, Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 87 should the difficulty of questions be tested? What is the minimum number of questions in a question bank? The IARCSC has thus far not been forthcoming with information regarding the exams, in part, they indicate, to avoid persistent attempts to unlawfully interfere in the recruitment process. However, given the robustness of the reforms, the lack of communicating them and the overall lack of transparency has been a missed opportunity to regain trust in the teacher recruitment process. 259. Second, the ‘single non-transferable application’ selection method will potentially result in a significant loss of qualified and willing candidates. Because this method does not select the highest scoring candidates in a category or ‘pool of vacancies’ (i.e. the twenty highest scoring chemistry teachers in province X), but rather the highest scoring candidate for a single vacant position. If the choice of the 118,544 applicants would be randomly distributed across vacancies, there would be no loss of valuable qualified candidates. However, the inherent risk in this method is that qualified candidates might concentrate on a few vacant positions, and therefore, many will not be selected. 260. This might appear as an unlikely hypothetical statistical scenario, but there could be structural reasons why choices are skewed (concentrated). If confident chemistry teacher applicants (those that received high grades at the TTC) apply on a single position at a popular school (a school with a good reputation), many will not be selected. In other words, not all the applicants with the highest scores on the chemistry teacher exam will be selected for the chemistry teacher positions. It is not clear how significant the impact of this loss will be, and a thorough analysis of the application database is required to assess this. However, the impact could be considerable in larger, dense urban areas with a considerable discrepancy of school quality (i.e. those with and without buildings). Figure 23: Example: ‘pooled selection’, 30 applicants apply for 3 positions - the best three are selected. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 88 Figure 24: Example: ‘single non-transferable application’, 30 applicants apply for 3 position, but only compete for the specific vacancy applied for. 261. Third, it is yet unclear to what extend the IARCSC was successful in promoting female teacher recruitment. According to the IARCSC, female candidates for the teacher entry exam received additional bonus points increasing their chances of being selected. In addition, teacher vacancies were gender specific. However, it is yet unclear to what extend the IARCSC was successful in increasing the number of women hired into the teacher workforce, especially in those provinces where the gender gap is most pronounced. 262. Estimates of provincial teacher recruitment GPIs, discussed in chapter 2, suggest that female teacher recruitment needs to be explicitly encouraged in some areas. Typically provinces with already low rates of female teachers, did not significantly improve their teacher GPIs through recruitment in the past years. Without a clear gender intervention in the latest IARCSC managed round of teacher recruitment, these provinces will likely continue to have abysmal rates of women selected for new positions. Targeted female recruitment is the most effective tool to increase women into the teacher workforce. However, it is unclear how effective IARCSC measures to promote female recruitment have been, and some officials interviewed for this study provided different account of what measures were taken69. 263. Last, the reforms in the teacher recruitment process have done little to facilitate rural teacher applicants. Although similar challenges existed in the old recruitment process, this is particularly important given the difficulty of finding qualified teachers in rural areas. The new recruitment process requires rural teacher applicants to travel two or three times to the provincial capital. First, to collect and submit an application document (although this was also made possible in many districts); second, to collect an exam ID card and register in the biometric database; and third, to complete the teacher entry exam. While this is little effort for urban applicants, the financial burden of traveling to the provincial capital many times (in some rural arras this takes more than a day) will have likely deterred rural candidates. 69 One IARCSC official indicated that no gender promotion measures were taken, because the commission was not allowed to distinguish between gender or promote female recruitment due to existing civil service and labor laws. However, another IARCSC official stressed that women did receive additional bonus points on the entry exam. A more thorough evaluation of the latest recruitment round is required to assess the effectiveness and impact of female recruitment promotion, especially in provinces where the current teacher GPI is low. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 89 264. Similarly, although generally laudable, the use of electronic media (website announcements, internet based applications) should not have entirely replaced old communication systems (hardcopy), as rural applicants often don’t have access to the Internet. Although to early to assess at this point (and indeed hardcopy systems were used in many cases for announcements), the initial data published by the IARCSC shows that in some rural provinces (e.g. Uruzgan) only half of the shortlisted candidates collected their exam ID and submitted their biometric data. Although it is to early to suggest this is the result of its implementation, the new recruitment reforms could have done more to facilitate rural recruitment. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 90 Recommendations – Teacher Recruitment The IARCSC should: SHORT-TERM: 265. Clearly describe the civil servant recruitment process, identify steps and regulations in order to improve transparency, accountability and increase awareness. This description should be published by the IARCSC and presented to its main stakeholder (the line ministries) as well as outside stakeholders (e.g. civil society). The IARCSC should clearly identify where, when, and for how long announcements, documents, results and complaints will be made available, and enable stakeholders to hold it accountable. 266. Develop standardized instructions for applicants to be shared during the announcement and application phases, which explain to applicants how processes work (application, examination, selection). The instructions should identify requirements, dates, time, and duration, as well as what mechanism exist for complaints and redress. 267. Thoroughly evaluate to ongoing recruitment round to better understand the validity and the fairness of the new exams, the impact of the chosen selection method (single non- transferable application), and its effectiveness of reaching target groups like women and rural applicants. Outcomes of the evaluation should be discussed with its main stakeholders (the line ministries) and ‘lessons learned’ to be implemented in the next round, should be identified. MEDIUM-TERM: 268. Develop transparent formulas/polices to integrate preferential treatment into the competitive recruitment process, to promote the recruitment and selection of target groups (e.g. women, rural applicants, applicants with certain subject qualifications), based on consultations with the line ministries. The Ministry of Education should: SHORT-TERM: 269. Clearly and transparently describe how the new teacher recruitment process works and communicate this to applicants and stakeholders. Prospective teacher applicants should be informed during the announcement and application phase of the process and regulations on applications (apply for one position only), biometric registration (the location and time period available), examination, and selection. 270. Identify target groups for preferential treatment in the competitive recruitment process, including women, rural teacher applicants, and applicants with scares subject specializations, and advise the IARCSC on how to advance their chances of being selected. 271. Use the IARCSC database on applicants, and exam results to develop a detailed and comprehensive overview of HR availability, and discrepancies in gender and (subject-) qualifications across provinces and districts. The overview should be used to inform teacher training needs, targeted recruitment policies and regional or gender based allowances. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 91 Photo 3: Teacher HR files (Nangarhar PED, 2015) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 92 5 Teacher Remuneration 272. This chapter examines Afghanistan’s teacher remuneration system. It describes tashkeel and non-tashkeel salary scales and allowances, and the civil servant pension system. It thereafter looks the competitiveness of the teacher remuneration package. This chapter ends with a discussion and recommendations. 5.1 Tashkeel teacher remuneration 273. Tashkeel teacher remuneration (i.e. a ‘regular teacher salary’) is regulated by national civil service laws, regulations, and scales. These rules and scales are the same everywhere in Afghanistan. There are no regional differences in salaries or allowances, nor are there formal institutionalized differences between men and women or urban and rural teachers70. Tashkeel teachers are paid twelve full monthly salaries per year, even though they are expected to work only nine months, and around 24 to 28 hours per week. Figure 25: P&G base salary in AFN - all BASTS & STEPS BAST 1 27.200 - 32.500 BAST 2 18.400 - 22.400 BAST 3 13.000 - 16.000 BAST 4 9.900 - 11.900 BAST 5 8.000 - 9.200 BAST 6 6.500 - 7.500 BAST 7 5.600 - 6.200 BAST 8 5.000 - 5.400 0 5000 10000 15000 20000 25000 30000 STEP 1 STEP 2 STEP 3 STEP 4 STEP 5 (Source: IARCSC, P&G Pay scale) 5.1.1 Pay & Grade ‘base’ salary-scale 274. ‘Base’ salaries of all civil servants (including teachers) are determined by the civil servant pay & grade (P&G) salary scale. This scale divides civil servant positions into eight grades or ‘basts’. Bast 1 is the highest grade, reserved for senior civil servant positions (director general level), and bast 8 is the lowest grade, reserved for support staff (e.g. cleaners, guards, 70 The CBE modality is an exception. However these are not ‘regular tashkeel teachers’, but teachers with a temporary contract and not funded though the recurrent budget. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 93 etc.). Basts are subdivided into five STEPS reflecting work experience, with STEP 1 the lowest STEP (no work experience), and STEP 5 the highest STEP (many years of work experience). A civil servant’s bast depends on his or her position, whereas a civil servant’s STEP depends on the years worked in that position. The P&G salary scale was introduced in 2009 (1388), with the pay and grade reforms, and ranges from 5,000 AFN (72 USD) monthly for the lowest ranked civil servants (cleaners, cooks, tea makers) to 32,500 AFN (462 USD) monthly for the most senior and experienced (provincial directors and director generals) see Figure 25. This scale identifies the civil servants ‘base’ salary exclusive of allowances, and before tax and pension deductions. 275. P&G base salaries of regular (qualified) tashkeel teachers (bast 5 and bast 6) range from 6,500 AFN to 9,200 ANF per month (93 USD to 131 USD). In fact, due to the years required to reach the final STEP in the promotion ladder, and the limited number of teachers hired in the past years, the P&G base salary of many teachers is either 107 USD (7,500 AFN for bast 6, STEP 5 teacher with a grade 14 degree) or 131 USD (9,200 AFN for bast 5, STEP 5 teacher with a grade 16 degree), see Figure 26. Figure 26: P&G monthly base salary in AFN, regular tashkeel teachers, by BAST and years of service usually grade 16 (university) graduates BAST 5 8000 +300 +300 +300 +300 usually grade 14 (TTC) graduates BAST 6 6500 +250 +250 +250 +250 6000 6500 7000 7500 8000 8500 9000 1st year 2nd year 3rd year 4th year +4 years (Source: IARCSC P&G Pay Scale) 276. However, for some teachers the base salary is lower or higher than the 6,500 to 9,200 AFN range. This includes ajeer teachers (usually under qualified teachers) in bast 7 and bast 8 receiving less than 6,500 AFN (see Figure 27), as well as some of the teachers who have yet to be graded according to the P&G scale and are currently in bast 3 or bast 4 (this group is still graded according regulations prior to the introduction of P&G). Figure 27: P&G monthly base salary in AFN, ajeer tashkeel teachers, by BAST and years of service BAST 7 5600 +150 +150 +150 +150 BAST 8 5000 +100 +100 +100 100 4800 5000 5200 5400 5600 5800 6000 6200 1st year 2nd year 3rd year 4th year +4 years (Source: IARCSC P&G Pay Scale) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 94 5.1.2 Professional and cadre allowances 277. In addition to the P&G base salary, civil servants receive a professional allowance dependent on their level of education. This allowance aims to incentivize the recruitment and retention of educated professionals into the civil service. The allowance is negligible for grades 12 and 14 educated teachers (e.g. High school and TTC graduates) but significant in proportion to the base salary for university undergraduates (grade 16) or higher educated teachers. The allowances ranges from 8 AFN per month for ‘grade 12’ educated teachers, to 6,000 AFN for teachers with a PhD (0.11 USD to 86 USD) see Figure 28. Figure 28: Monthly professional allowance in AFN, by education qualification of teacher PhD +6000 AFN Master +4000 AFN Bachelor +2000 AFN Grade 14 +12 AFN Grade 12 +8 AFN 0 1000 2000 3000 4000 5000 6000 (Source: Payroll Registration Unit) 278. Teachers that teach tertiary education (classes above grade 12), here referred to as ‘lecturers’, receive an additional cadre allowance. Lecturers work at TTCs, TVET institutes, and some literacy and Islamic Education institutes. Lecturers also work at public universities administered by the Ministry of Higher Education (MoHE). However, none of the regular teachers in general education receives a cadre allowance. Cadre allowances were introduced to incentivize recruitment and retention of professionals in academia. The allowance is determined by the academic title (dependent on the grade of tertiary education taught, as well as the number of academic publications), and ranges from 8,000 AFN to 12,000 AFN (114 USD to 171 USD). Please see Figure 29 for an example of cadre allowances for TTC lectures, and see appendix 2 for more details. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 95 Figure 29: Cadre allowance TTC lecturers, by academic title Showondae 12,000 AFN Showonmal 10,500 AFN Showanyar 9,000 AFN Shohyalae 8,000 AFN 0 3,000 6,000 9,000 12,000 (Source: Payroll Registration Unit) 5.1.3 Food and transportation allowances 279. All civil servants, including all tashkeel teachers (both regular and ajeer) receive an additional food (or lunch) allowance. This lunch allowance is a fixed 30 AFN (0.43 USD) per day for all civil servants no matter their grade or seniority. This food allowance is calculated based on 22 workdays per month (even though most teachers are expected to work 6 days per week or almost 26 workdays per month). As a result, all tashkeel teachers receive a 660 AFN (9.41 USD) monthly lunch allowance on top of their base salary and professional/cadre allowance. 280. Tashkeel teachers do not receive a transportation allowance (to compensate for their daily commute to and from work). Unlike the lunch allowance, not all civil servants receive a transportation allowance. In the case of the Ministry of Education, only staff of the Ministry’s departments in Kabul and the Kabul City PED receive a daily transportation allowance. This allowance is fixed at 80 AFN per day and 1,760 AFN per month (1.14 USD and 25 USD). However, the payroll registration unit indicated that some PEDs received permission from MoE’s leadership to pay transportation allowances to their staff, and individual exceptions to this rule exist. 5.1.4 Deductions: Pension and Tax 281. Tashkeel teachers pay an eight percent pension contribution. This eight percent is deducted from their total salary (base salary as well as professional and cadre allowances), and is the most significant deduction of a typical teacher salary. For a small portion of tashkeel teachers, only a three percent pension contribution is deducted from their salaries. This group of teachers has not been re-graded according to P&G yet (7,931 teachers are not yet re-graded according to the MoE Finance Department). 282. Teachers typically pay around 1% tax (effective tax rate) on their salary and allowances. Tax is calculated over their base salary, professional and cadre allowances, as well as their food allowance. The first 5,000 AFN (71 USD) of a monthly salary is not taxed. Monthly salaries between 5,001 and 12,500 AFN (178 USD) are taxed at 2%, salaries from 12,501 AFN to 100,000 AFN are taxed at 10%, and thereafter the tax rate is 20%. See a calculation example in Figure 30. For more examples of gross and net remuneration calculations, see Table 12. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 96 Figure 30: Example of pension and tax deductions over a BAST 5 STEP 5 teacher gross salary Pension 8% over 11.200 AFN GROSS salary 9.200 +2.000 +660 Tax 0% over 5.000 AFN 2% over 6.860 AFN 0 2000 4000 6000 8000 10000 Base Salary Professional Allowance Food Allowance (Source: Staff Calculations) 5.1.5 Deductions: Absenteeism 283. Teacher salaries are deducted if a teacher exceeds the maximum number of leave days. Tashkeel teachers are expected to work nine months per year. Within those nine months, they are allowed to take up to 30 leave days (20 days for ‘sick leave’, and 10 days for ‘urgent leave’). Non-teacher civil servants are expected to work 12 months per year but receive 20 additional ‘annual/leisure’ leave days). If a teacher exceeds his or her leave limits, both the base salary and professional and cadre allowances are deducted. The deduction is calculated based on 22 workdays per month. One additional leave day (exceeding the limit) results in a 1/22th reduction (-4.5%) of the total monthly salary. However, the lunch allowance is always deducted from the teacher’s monthly salary, even if the leave/absenteeism is within the regular leave limits. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 97 Table 12: Examples of net salary calculation (regular and non-regular) teachers (fictitious examples) Regular Non-regular Teacher TEACHER 1 TEACHER 2 TEACHER 3 TEACHER 4 Dari Math Biology Subject Chemistry Teacher Teacher Teacher Teacher Teaches class Grade 11 Grade 4 Grade 3 Grade 10 Age 56 25 48 31 Sex Female Male Male Female Province Uruzgan Kabul (City) Nuristan Bamiyan Contract Regular tashkeel Regular tashkeel Tashkeel Ajeer Outside-haquzama Degree Grade 14 Grade 16 Grade 12 Grade 14 From TTC University High school TTC Has worked as a 24 years 4 years 18 months 20 years teacher for Bast 6 5 8 N/A STEP 5 5 2 N/A Base Salary 7,500 9,200 5,100 5760 Professional 12 2,000 8 N/A Allowance Cadre allowance N/A N/A N/A N/A Gross salary for 7,512 11,200 5,108 N/A pension Food allowance 660 660 660 ? Transportation N/A N/A N/A N/A allowance Gross salary for 8,172 11,860 5,768 5760 tax 8% Pension 600.96 896 408.64 n/a deduction 0% tax up to 0 0 0 0 5,000 AFN 2% tax after 63.44 137.2 15.36 15.2 5,000 AFN NET monthly 7,508 10,827 5,344 5,745 salary AFN 1 USD = 70AFN $ 107 $ 155 $ 76 $ 82 (Source: Staff Calculations) 5.2 Non-tashkeel teacher remuneration 284. Salaries of non-tashkeel teacher contracts (i.e. teachers with outside-haquzama or bilmakta, but also regular tashkeel teachers with an additional inside-haquzama contract) are regulated nationally through specific bills (i.e. the Bilmakta-bill, see annex 7, and the Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 98 Haquzama-bill, see annex 8). Haquzama remuneration is the same everywhere in Afghanistan, and does not differentiate between male, female, urban or rural teachers. Haquzama teachers usually receive a nine-month contract and nine full-time salaries. However, these contracts can also be for shorter periods (e.g. haquzama contracts are sometimes used to cover for the three- month maternity leave of a regular tashkeel teacher). 5.2.1 Haquzama remuneration 285. The salaries of haquzama contracts are calculated based on a fixed hourly fee, depending on the education level of the contracted teacher. Fees range from 60 AFN (0.86 USD) per hour for ‘grade 14’ educated contract teachers; 70 AFN (1 USD) per hour for grade 16; and 80 AFN (1.14 USD) per hour for contract teachers with a master’s or PhD. Remuneration is the same for both outside- and inside-haquzama teachers. 286. Haquzama contracts for general education teachers (in primary and secondary education) are always for 24 hours per week, and thus one full-time teacher equivalent (1 FTE). Haquzama salaries are calculated based on the hourly fee, multiplied by 96 hours (24 hours in each 4 weeks) per month. Monthly salaries thus range from 5,760 AFN (82 USD) for grade 14 educated contract teachers; 6,720 AFN (96 USD) for grade 16 educated; and 7,680 AFN (109 USD) for contract teachers educated above grade 16, see Figure 31. There are also haquzama contracts for non-teaching and part-time teaching staff, such as principals and librarians. These are not full-time equivalent contracts and thus their salaries are calculated using different monthly hours. Figure 31: Inside- and outside-haquzama teacher salaries for a 96 hours per month contract Master 7680 Grade 16 6720 Grade 14 5760 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 (Source: Haquzama Bill, interviews) 287. Haquzama contract teachers do not receive professional or cadre allowances, nor do they receive lunch allowance. Haquzama contracts do not include transportation allowances, nor pension deductions. Teachers with haquzama contracts do need to pay tax over their income. Tax brackets are the same as for regular tashkeel teachers, and inside-haquzama payments should be considered part of a teacher’s total income to calculate tax71. 71 A teacher’s regular tashkeel salary is often paid separately from his or her inside-haquzama salary. Tax is mostly calculated on the spot (when filling the salary payment sheet) and thus tax over many inside-haquzama contracts is calculated as if it was a different individual. However, one MoE officials indicated that larger PEDs (e.g. Kabul City) regular and haquzama salaries are often jointly processed and collectively taxed. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 99 5.2.2 Bilmakta-ajeer teacher remuneration 288. Bilmakta-ajeer teacher contracts are the least common within the pool of teachers. Bilmakta salaries are not fixed. The bilmakta regulation indicates that salaries need to be negotiated within a range from 5,000 AFN (71 USD) to 13,000 AFN (185 USD), see Figure 32. This range is equivalent to the P&G base salary of bast 8 STEP 1, to bast 3 STEP 1 civil servant. Bilmakta contracts can be used to hire both professional and support staff, but in the case of teachers, these contracts are mostly used for under-qualified employees. Bilmakta-ajeer teachers do not receive professional or cadre allowances; do not receive lunch allowances; are not deducted for pension contributions; but do need to be deducted for tax. According to MoE officials, bilmakta contract teachers usually receive a nine-months contract and salaries. Figure 32: Salary range of Bilmakta employees Negotiated up to Bilmakta Minimum 5000 13.000 0 2000 4000 6000 8000 10000 12000 14000 (Source: interviews) 5.3 Teacher Pensions 289. Pensions are regulated by the civil service law and regulations, and pensions are the same for teachers as well as other civil servants. Only regular tashkeel teachers, tashkeel ajeer teachers, and MoE tashkeel staff pay into – and later receive pensions. Civil servants with haquzama or bilmakta contracts do not pay into, and do not build up pension. The size and duration of pension payments depend on three key factors: the last earned salary, the duration of public service, and the ‘reason of retirement’ and pension-payout. 290. There are five different conditions under which a teacher can retire and/or receive a pension: • A teacher retires when reaching the age of 65 or after 40 years of public service; • A teacher chooses to retire ‘early’, at the age of 55 or after 25 years of public service; • A teacher dies before reaching the age of 65 or 40 years of public service; • A teacher falls sick or is disabled and permanently unfit for work; • A teacher’s contract is terminated voluntarily (own will) or involuntarily (fired). 291. The size of pension payment depends on the duration of service and the size of the last monthly salary: • With less then 10 years of public service, the teacher will receive 2.5 times the last monthly salary per completed year, and the total amount will be disbursed in a single lump-sum; • With 10 years of public service, the teacher will receive 40% of his last monthly salary, the amount will be calculated and disbursed annually; • For each additional year in public service, the teacher will receive an additional 2% of his last monthly salary, calculated and dispersed annually; • With 40 years of public service, the teacher will receive 100% of his last monthly salary, calculated and disbursed annually. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 100 292. In addition, in the case of death, permanently unfit for work due to illness or disability, or when reaching the full retirement age / years of service, a retiree receives ‘Ikramya’ (a golden handshake). The retiree’s ‘golden handshake’ is ten times the last monthly salary paid at once . Furthermore, if a teacher dies at his or her work location, i.e. at school, a full pension is paid out to his/her widow(er), no matter how long the diseased teacher has served. 293. Last, the pension is paid for the remainder of the retiree’s life. If the retired teacher is survived by his/her widow(er), pension continues to be paid until the death of his/her widow(er). If both parents in the teacher’s family die while their children are still under-aged and unmarried, the pension will be paid until the children reach adulthood. In the latter case, a trustee is appointed, usually a brother of the retiree, to receive the annual pension payment (claims of survivor benefits being paid for the remainder of the widow(er)’s life were disputed and an IARCSC official indicated that survivor benefits were only paid for 3 years.). 5.4 Competitiveness and real income 294. Private school teacher salaries are generally not perceived as better than public school teacher remuneration. Both teachers and private school managers interviewed for this study indicated that salaries of private school teachers are usually similar or lower than the salaries of public school teachers. There are more than 1,200 private schools in Afghanistan, employing more than 20,000 teachers. Half of those private school teachers work in Kabul, the other half work mostly in larger urban centers around the country72. This makes the private sector a suitable alternative for prospective teachers, primarily in urban centers. 295. While some teachers in the private sector earn more than their public sector colleagues, most earn the same or less. The Ministry of Education does not regulate private school teacher salaries and there are no laws, regulations or fixed salary scales73. Teacher salaries depend on school entry fees (i.e. expensive private schools having better salaried teachers), education qualifications of the teacher, and market dynamics (i.e. supply and demand of qualified teachers). As a result, teachers with rare subject specializations can earn more in the private sector than at public schools. However, private sector jobs are generally not better paid, do not provide pension or significant benefits, and are nor perceived as more competitive than public sector teacher positions. 296. Current salaries of public school teachers, working 24 to 28 hours per week, are not perceived as sufficient to sustain a family. For male teachers, a second job (e.g. as a shopkeeper, or private tutor) is the rule rather than the exception, according to teachers interviewed for this study. In Kabul, teachers typically prefer to work the early morning shifts, so that they keep the afternoon for secondary employment. For female teachers, the range of options for finding culturally accepted employment is narrower, and second jobs are less 72 These statistics are based on EMIS data. Estimates of the size of the private education sector vary. One private school director estimated the total number of private schools in the country around 2,100 of which only a third in the capital. 73 The salary and other terms and conditions related to payment (including overtime if both parties agreed) are mentioned in the agreement between a teacher and a private school. In most cases, there is no pension or retirement allowance for private schools’ teachers. The MoE does not control or interfere in the salary payment of private schools teachers, however, if private schools delay the payment of their teachers, MOE then interfere to resolve the problem. Private schools established three unions (national level, Kabul City level, and Provincial level) that facilitate the affairs of private schools with the MOE, other government organizations, government and private schools, NGOs, and other stakeholders. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 101 common. Female teachers interviewed for this study indicated that in urban areas, female public teachers often also work part-time in private schools. Many retired tashkeel teachers also continue to work as outside-haquzama teachers, despite receiving full pension. “A generation ago, a teacher could buy a house and sustain a family with nine children. Nowadays, my salary is not enough to cover the rent, let alone to buy food�.74 297. In 2015 (1394), teachers in the capital Kabul protested the rising cost of living and the perceived lack of adequate remuneration75. They demanded better pay, a more significant promotion/career trajectory, and the distribution of land plots to teachers76. None of these demands have been met since. Only one proposal has been prepared to reform promotion opportunities for teachers (as of late 2017, this proposal is under revision with the Ministry of Justice). 298. Since the introduction of the P&G scale in 2008, civil servant (and teacher) salaries have not been adjusted for inflation, despite significant year-on-year inflation in the same period. Consumer prices have risen 34% since 2009, meaning real civil servant (and teacher) salaries (adjusted for inflation) have dropped with a quarter77, see Figure 33. Figure 33: Consumer Price Index and real wages 2010-2016 160 120% 133 131 134 140 127 118 100% 120 110 99 80% 100 75% 80 60% 60 40% 40 20% 20 0 0% 2010 2011 2012 2013 2014 2015 2016 CPI (2010=100) real income % (Source: International Monetary Fund) 299. Some MoE officials interviewed for this study stressed that regular civil servants, rather than teachers suffered the most from the depreciation of their income, since they are expected to work a 40-hour workweek and are less able to generate income from a secondary job. However, the drop in real income has further added to the perception that a teacher’s salary is not sufficient for cost of living, but rather a stable part-time job that needs to be supplemented with additional income. 300. The latest round of teacher and non-teacher civil servant recruitment in early 2018 suggests that the teacher profession remains relatively attractive for job seekers. Nation-wide, 118,544 applicants were shortlisted for 8,000 teacher positions (almost 15 applicants per 74 Teacher interviewed for this study. 75 Teacher Strike Spreads Across Kabul. Tolo News, June 3, 2015. 76 The distribution of publicly owned land to civil servants is common practices in Afghanistan. 77 International Monetary Fund, International Financial Statistics and data files, retrieved from the World Bank website. In the same period, the Afghani lost almost 33% of its value against the US Dollar. The teacher P&G base salary initially ranged from approximately $156 USD to $192 USD. This has drop since to $108 USD to $131 USD. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 102 position). Another 106,778 applicants were shortlisted for 9,700 non-teacher civil servant positions (on average 11 applicants per position). The considerable number of applicants in both cases appears to suggest that despite a significant drop in real income in the past years, teacher and civil servant remuneration remains competitive compared to other employment sectors. 5.5 Discussion and Recommendations 301. The current remuneration scale and regulation does not enable incentivizing policy priorities. Within current regulations, there is no option to deviate from salary standards given exceptional conditions (besides the professional and cadres allowances to incentives teacher qualifications). Salaries are the same for everyone and everywhere. Remuneration is equal across provinces and districts, and the same for men and women, rural and urban teachers. Even though there are considerable regional variations in economic development, education, cultural norms, security and urbanization in Afghanistan, the Ministry is limited to a one-size fits all remuneration scale. 302. Rural schools across Afghanistan are struggling to find willing and qualified candidates for their positions, and qualified female candidates are exceptionally hard to find outside of urban centers and in most southern and eastern provinces78. Yet, at the moment, a female teacher in Deh Rawud, rural Uruzgan, receives the same remuneration package as a male teacher in Kabul city, and the MoE has no tools to incentivize female, regional or rural recruitment. 303. Other public sectors (e.g. public health) do allow for regional, gender and hazard allowances. And other line ministries use the bilmakta modality to recruit professionals in areas of scarcity. The bilmakta salary range is limited (up to $185 USD), but it allows for some flexibility in remuneration and thus to incentivize recruitment in areas of human resources scarcity. The BPHS/EPHS health regulations give health providers considerable flexibility in paying regional, rural and female incentives, allowances and bonuses. 304. Education programs funded through the development budget (i.e. not regulated by civil service regulations) and off-budget modalities have paid considerable stipends to regular female teachers willing to work in provinces where there are almost none. In one example given by an interviewee, female teachers and their (teacher) spouses were able to earn between four and six times their regular salaries if they were willing to work in Paktika province. 305. Given the considerable variations in human resource constraints, the Ministry lacks the necessary remuneration tools to provide equitable education services to all children in Afghanistan. The development of adequate and flexible remuneration tools to address education services disparities will require two things. First, it requires a much more thorough understanding of the human resource constraints and disparities in the country. Decentralized recruitment and management of applicant data have prevented such an assessment, but the new centralized database will provide an initial input into understanding this issue. And second, 78 There has been hitherto no centralized collection of data on applicants (i.e. location, gender, and qualification) and it is thus not possible to asses the degree to which regions differ in terms of the availability of qualified teacher applicants. However, the ongoing IARCSC effort to recruit new civil servants across the country will provide an abundance of detailed data on this issue. Furthermore, most officials interviewed for this study stress that qualified teacher candidates are in abundance in most urban centers, but very scarce in remote and insecure regions. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 103 strict regulations, transparency and accountability will be required to address the inherent risks of flexible remuneration and allowances, and to avoid abuse. 306. Flexible remuneration tools increase the risk of abuse and they are not necessarily effective in addressing the issues they are intended for. While the lack of flexibility currently prevents the MoE to deliver equal services across the country, more flexibility does not automatically solve this equity issue and all remuneration reforms should consider current limitations administrative capacity and oversight. 5.5.1 Salaries are static and based on level of education 307. Salaries change only the first four years of a teacher’s career, after which the salary remains unchanged for 36 years, according to the P&G scale. In the current scale and promotion policy, a typical teacher (either bast 6 or bast 5) sees his or her base salary increase around 3.8% each year, for the first four years. Note that the annual increase in salary, based on promotion, is almost equal to the annual increase of consumer prices. After four promotions in four years, moving from STEP 1 to STEP 5, the base salary has increased by around 15%. In the remaining 36 years of his or her career as a civil servant, there is nothing the teacher can do to change this salary beside getting a better degree (which is financially impossible for most teachers), working double shifts (through a haquzama contract) or changing his or her job. 308. Salaries are almost entirely dependent on the teacher’s level of education. In the pool of qualified teachers with more than four years of work experience, there are effectively two fixed Net salaries: 7,508 AFN (8,172 gross including lunch allowance) and 10,827 AFN (11,860 gross including lunch allowance), or 107 USD and 155 USD respectively. These salaries are unrelated to workload, performance, or responsibilities. In the fictitious example given in Table 12 “TEACHER 1� (female) has been teaching chemistry to grade 11 students in rural Uruzgan for 24 years. “TEACHER 2� (male) has been teaching Dari to grade 4 students in Kabul City for a little over four years. Yet, according to the current teacher remuneration system, “TEACHER 2� earns almost 50% more than “TEACHER 1� because he has a university degree rather than a TTC degree. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 104 Recommendations – Teacher Remuneration The Ministry of Education should: 309. Participate in the national dialogue on public sector remuneration led by the IARCSC and Ministry of Finance officials. This dialogue should focus on resolving the challenges of regional pay disparities, declining real salaries, and integration of incentives in the remuneration system. Participation in this process should seek to address these challenges within the medium-term fiscal envelope. The creation of a Civil Service Pay Commission, led by MoF and IARCSC, could facilitate policy dialogue on remuneration reforms. A teacher pay taskforce within the commission could look more in depth into specific challenges affecting teacher management. Photo 4: Teachers applying for their STEP promotion (Kabul City) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 105 6 Teacher Costs 310. This chapter discusses teacher costs. It first compares teacher salary expenditures to other MoE expenditure categories, and calculates per student and per teacher expenditures. It thereafter estimates teacher costs, looks at historical developments and regional differences, and concludes with forecasting teacher costs until 2025. Please note that all USD amounts presented in brackets “(USD)� are indicative only, converted to a fixed 70 AFN per USD rate, unless otherwise indicated. 6.1 MoE Expenditures 311. Teacher base salaries alone are responsible for half of all MoE expenditures. In 2016, the MoE spent almost 18 billion AFN (256 million USD) on teacher base salaries. In addition the MoE spent 4.3 Billion AFN (61 million USD) on non-teacher base salaries, 3.6 billion AFN (52 million USD) on ‘overtime’ payments, and 2.4 billion (34 million USD) on other wage allowances (including 0.68 billion AFN on professional and cadre allowances, and 1.6 billion AFN on food allowances). The MoE wage bill absorbs 78% of all MoE expenditures, leaving 6% for operations and maintenance expenditures, and 16% for development investments (e.g. infrastructure and large procurement contracts) see Figure 34. Figure 34: Overview of MoE expenditures per category 2016 (Source: AFMIS) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 106 313. MoE total expenditures increased by 64% (nominal) from 2010 to 2014, from 23.1 billion AFN (505 million USD, 2010) to 38 billion AFN (634 million USD, 2014) and thereafter plateaued, see Figure 35 (the 1391 fiscal year only consisted of 9 months, and the noticeable dip in expenditures does not reflect a dramatic change of trends). Similarly, teacher base salaries expenditures rose significantly between 2010 and 2014 and remained nearly unchanged in the following years. Teacher base salary expenditures increased by 60% between 2010 and 2016, from 11.2 billion AFN (245 million USD, 2010) to 17.9 billion AFN (256 million USD). The considerable increase in base salary expenditures is explained by two factors in this period, the increase in the number of tashkeel teachers (+13.3%) and the increase in average base salary expenditure per tashkeel teacher (42%, as will be discussed in section 6.2). Figure 35: MoE expenditures by category (billions AFN) 2010-2016. Note: the 1391/2012 fiscal year only consisted of nine months. 5.8 35 30 2.2 2.4 3.6 25 4.0 4.3 20 1.7 17.3 17.3 17.9 2.2 9 15 1.0 month 15.5 2.8 13.6 FY 11.2 10.9 10 05 1389 1390 1391 1392 1393 1394 1395 2010 2011 2012 2013 2014 2015 2016 TEACHER SALARIES NON-TEACHER SALARIES OVERTIME 21 ALLOWANCES O&M DEVELOPMENT BUDGET (Source: AFMIS) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 107 314. MoE expenditures compared to Afghanistan’s nominal GDP declined in the 2010-2016 period. Nominal GDP increased from 729 billion AFN (15.9 billion USD, 2010) to 1,341 billion AFN (19.2 billion USD), an increase of 84%. Total MoE expenditures fluctuate between 3.09% and 3.29% of GDP until 2014 and thereafter decreased. In 2015 and 2016 MoE expenditure did not increase (even decreased somewhat) while GDP continued to grow at around 8% annually, see Figure 36. Similarly, MoE wages and salary expenditures fluctuated between 2.34% and 2.48% of GDP until 2014, and thereafter decreased to 2.19% in 2016. Figure 36: MoE expenditures as percentage of GDP 2010-2016 3.20% 3.29% 3.17% 3.09% 3.01% 2.79% 2.39% 2.44% 2.48% 2.34% 2.33% 2.19% 2.19% 1.74% 1389 1390 1391 1392 1393 1394 1395 2010 2011 2012 2013 2014 2015 2016 Wages and Salaries Total MoE expenditures (Source: AFMIS, IMF) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 108 315. While nominal MoE expenditures have increased considerably, the relative distribution of expenditures per category, within the MoE, did not fluctuate significantly, see Figure 37. Teacher base salaries absorbed between 46% and 50% of total MoE expenditures (2010-2016), non-teacher base salaries between 12% and 13%, O&M between 6% and 8%, and development expenditures between 13% and 19%. However, overtime expenditures, as a percentage of total MoE expenditures, rose significantly from only 4% in 2010 to 12% in 2013, and thereafter leveled around 10%. In absolute terms, MoE overtime expenditures rose by 260% between 2010 and 2016, from 1 billion AFN (22 million USD, 2010) to 3.6 billion AFN (52 million USD). Figure 37: MoE total expenditures (millions AFN) 2010-2016 and annual expenditures by code in percentages 36,678 36,246 35,323 33,705 19% 16% 17% 17% 26,993 7% 6% 6% 8% 5% 7% 6% 21,810 17% 6% 10% 21,880 11% 10% 18% 7% 12% 7% 13% 12% 12% 12% 7% 7% 8% 12% 9% 6% 49% 13% 10% 47% 49% 4% 46% 12% 50% 12% 49% 50% 1389 1390 1391 1392 1393 1394 1395 2010 2011 2012 2013 2014 2015 2016 TEACHER SALARIES NON-TEACHER SALARIES OVERTIME 21 ALLOWANCES O&M DEVELOPMENT BUDGET (Source: AFMIS) 6.1.1 Overtime expenditures 316. The magnitude of MoE’s expenditures on overtime is significant. In 2016, for every 6,000 AFN spent on salaries (both tashkeel teachers and non-teacher tashkeel employees) the MoE spent an additional 1,000 AFN on overtime. With more than 250,000 tashkeel employees in 2016, this suggests that at least 41,000 additional FTEs were being paid through overtime expenditures. Moreover, given that overtime payment scales are lower than regular base salaries (see section 5.2), and a typical haquzama teacher receives only a nine-month salary, the number of FTEs paid through overtime expenditures is likely significantly higher. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 109 317. In fact, actual 2016 MoE expenditures on the ‘overtime’ budget line are sufficient to cover between 51,800 and 60,500 additional haquzama FTEs79. This is significantly higher than the 35,824 haquzama contracts reported by the MoE budget unit in late 2017. While, the ‘overtime’ expenditure code covers not only haquzama teachers but also overtime payments to MoE managerial staff, overtime expenditures cannot simply be converted into haquzama FTEs80. However, this calculation stresses the importance of accurate data on haquzama contracts and overtime payments. Moreover, it suggests that the MoE datasets on haquzama contract are likely underestimating the actual number of haquzama FTEs. 6.1.2 Per student expenditures 318. In 2016, the MoE spent on average 4,050 AFN (58 USD) per enrolled student. This includes wages and salaries, overtime, O&M and development expenditures. Expenditure per enrolled student increased by 33% between 2010 and 2014, and thereafter decreased somewhat, reflecting continued student growth (though marginal) while expenditures plateaued, see Figure 38. Figure 38: Total MoE expenditures per enrolled student 2010-2016 (USD corrected for annual average currency conversion) 4.119 AFN 4.265 AFN 4.015 AFN 4.050 AFN 3.669 AFN 3.833 AFN 3.202 AFN 71 USD 71 USD 63 USD 58 USD 75 USD 72 USD 70 USD 1389 1390 1391 adjusted 1392 1393 1394 1395 2010 2011 2012 2013 2014 2015 2016 (Source: AFMIS and EMIS) 79 This calculation divides the amount spent on ‘overtime’ by the annual (9 month) remuneration of a qualified (grade 14 and 16 educated) haquzama teacher. 80 In 2018, AFMIS expenditures codes for MoE administrative staff overtime payment and haquzama salaries were separated. This will allow for more accurate estimates in the future. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 110 319. Provincial MoE expenditures per student fluctuate year-on-year due to significant fluctuations in development and overtime expenditures. To more accurately reflect provincial trends, average annual expenditure per student across seven years is calculated. Between 2010 and 2016, the MoE spent on average 3,904 AFN (56 USD, current) per student. Provincial expenditures (annual average 2010-2016) varied from 2,274 AFN in Badghis and 2,432 AFN in Paktika, to 5,077 AFN in Zabul and 6,821 AFN in Panjshir, see Map 7. That is, average annual expenditures per student in the past seven years, was almost three times higher in Panjshir than in Badghis. Map 7: Total MoE annual expenditures per enrolled student by province, 2010-2016 average81 (AFN in color, USD in digits, fixed exchange rate 70AFN = 1 USD) 57 47 43 53 45 42 50 58 < 2,700 AFN 42 97 58 2,701 – 3,000 AFN 32 55 70 61 3,001 - 3,300 AFN 55 49 57 3,301 – 3,600 AFN 63 40 3,601 – 4,100 AFN 37 57 46 37 4,101 – 4,400 AFN 43 >4,400 AFN 39 41 46 35 55 N/A KABUL CITY 73 56 NATIONAL AVERAGE 54 49 52 (Source: AFMIS and EMIS) 320. The significant variations in per student expenditure cannot be explained by variations in development investments (e.g. school construction) alone. Development investments do alter per student expenditures in some provinces, but when analyzing per student wages and salaries expenditures, a very similar trend arises. 321. In 2016, the MoE spent on average 3,179 AFN (45 USD) on wages and salaries per enrolled student. Wages and salary expenditures per student vary significantly from 1,804 AFN (26 USD) in Uruzgan, to 6,317 AFN (90 USD) in Panjshir. However, year-on-year expenditures fluctuate somewhat (due to overtime expenditure fluctuations). To equalize provincial fluctuations, multi year (2010-2016) average expenditures per enrolled students are 81Average wages and salary expenditures per student in Kabul City have been omitted, because they include Kabul City PED employee salaries as well as central line ministry salaries. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 111 calculated82. The MoE spent on average 3,004 AFN (43 USD, current) on wages and salaries per student (between 2010-2016). Average wages and salary expenditures vary from 1,780 AFN (25 USD) in Badghis, to 5,286 AFN (76 USD) in Panjshir, see Map 8. Thus, the MoE spent three times more on wages and salaries per student in Panjshir than in Badghis, on average between 2010-2016. 322. In addition to PTRs, there are two factors determining per student expenditures on wages and salaries. Provinces with low wages and salary expenditures per student (Badghis, Paktika, Daikundi, Uruzgan) typically have a combination of low overtime expenditure rates and low average teacher costs (see section 6.2). Provinces with high wages and salary expenditures per students typically have high rates of overtime expenditures and high average teacher costs, in addition to low PTRs. Map 8: Wages and Salaries annual expenditures per enrolled student by province, 2010-2016 average83 (AFN in color, USD in digits) 51 42 37 47 41 39 40 54 < 2,200 AFN 37 76 38 2,201 – 2,500 AFN 25 46 62 52 2,501 - 2,800 AFN 46 42 42 2,801 – 3,100 AFN 52 36 3,101 – 3,400 AFN 33 47 36 29 3,401 – 3,700 AFN 35 > 3,700 AFN 34 35 31 28 44 N/A KABUL CITY 48 43 NATIONAL AVERAGE 35 40 37 (Source: AFMIS, EMIS) 6.2 Teacher Costs 323. This section estimates actual teacher costs using two different methods. The first method analyses tashkeel teacher expenditures (salaries and allowances) against the number of registered tashkeel teachers. A problem with this method is that not all tashkeel teachers are 82 This calculation adjusted 1391 expenditures (a 9-month fiscal year) in AFMIS to reflect a 12-month equivalent, to avoid underestimating average annual expenditures. 83 Average wages and salary expenditures per student in Kabul City have been omitted, because they include Kabul City PED employee salaries as well as central line ministry salaries. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 112 labeled correctly, and not all tashkeel teacher expenditures are accurately coded. The second method calculates total MoE wages and salary expenditures against the number of teacher FTEs. A key problem with this method is the lack of accurate data on teacher FTEs (i.e. haquzama contract data). 6.2.1 Salary expenditures per tashkeel teacher 324. In 2016, the MoE spent 7,455 AFN (106 USD) on teacher base salaries per tashkeel teacher per month, and 89,457 AFN (1,278 USD) per year. This amount is calculated by dividing total 2016 MoE expenditures on teacher salaries (AFMIS codes 21107 to 21109), by the total number of tashkeel teachers in the payroll registration database. The amount reflects expenditures on teacher base salaries only (allowances not included). A note of caution is required here. AFMIS expenditure codes for teacher salaries are not always accurately applied by PEDs, and not all teachers are accurately labeled as teachers in the payroll registration database. 325. Monthly salary expenditures per teacher appear somewhat low, compared to the pay scale of regular (qualified) teachers, between 6,500 AFN and 9,200 AFN (93 USD and 131 USD). A likely explanation for this is the relatively high number of under qualified teachers in the tashkeel pool (43% of tashkeel teachers has no ‘grade 14’ or higher qualification). The pay scale of under qualified teachers ranges between 5,000 AFN and 6,200 AFN (71 USD and 89 USD). 326. In addition, the MoE spent on average 645 AFN per tashkeel teacher on food allowances, and 281 AFN on professional and cadre allowances in 2016, putting average monthly expenditures (base salary plus allowances) per tashkeel teacher at 8,381 AFN, and annually at 100,564 AFN (120 USD and 1,437 USD), see Figure 39. Figure 39: Average salary and allowances expenditure per tashkeel teacher 2016 7.455 645 281 5000 5500 6000 6500 7000 7500 8000 8500 Base Salary Foor Allowance Professional & Cadre Allowances (Source: AFMIS, Payroll Registration Database) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 113 327. Average (base) salary expenditures per tashkeel teacher rose by 42% between 2010 and 2016. This is a significant increase given that the P&G pay scale was not adjusted in this period. This could partly be explained by teacher promotion within and between basts. Teacher base salaries increase by 15% within basts after the first four years of employment. Moreover, the base salary of a qualified teacher (bast 6) is 30% higher than the salary of an under qualified teacher (bast 8). However, the significant increase in the number of qualified teachers (+22% between 2011 and 2016), and within bast promotion cannot fully explain how base salary expenditures have risen by 42% between 2010 and 2016. Promotion alone could account for only 10% to 20% of the expenditure increase. It can also not explain the notable jumps (see Figure 40, notable increases between 2010-2011 and 2013-2014) in average base salary expenditures per tashkeel teacher. Figure 40: MoE base salary expenditures per tashkeel teacher 2010-2016 100,000 16,000 89,457 90,000 84,854 84,879 14,000 73,501 75,530 76,643 80,000 12,000 70,000 63,054 60,000 10,000 50,000 8,000 7,071 7,073 7,455 40,000 6,125 6,294 6,387 6,000 30,000 5,255 4,000 20,000 10,000 2,000 0 0 1389 1390 1391 1392 1393 1394 1395 2010 2011 2012 2013 2014 2015 2016 Annually Monthly (Source AFMIS and Payroll registration database84) 328. One possible explanation is the gradual implementation of pay and grade (P&G) reform. Many teacher salaries were not adjusted according to the current P&G salary scale in 2010. Adjusting teacher salaries in the 2010-2016 period might have caused the significant increase in base salary expenditures. More rigorous analysis is needed to understand the 42% increase in per tashkeel teacher base salary expenditures. Again it needs to be noted that per tashkeel teacher expenditures might be somewhat unreliable due to inconsistent expenditure coding and teacher labeling. 329. Regional differences 330. Significant regional differences exist between average base salary expenditure per tashkeel teacher. Average base salary expenditures per tashkeel teacher were analyzed for 35 PEDs over a seven-year period (2010-2016). The analysis resulted in clearly discernable trends, but also appeared to reveal considerable errors by individual PEDs (in coding and labeling). As such, results of this analysis should be used with caution. 84Due to a switch in calendars, Afghanistan’s 1391 fiscal year covered only a 9-month period. Annual average expenditures per teacher for 1391 have been adjusted to reflect a 12-month period, so as to be comparable to other years. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 114 331. In 2016, the Kabul City PED spent on average 8,650 AFN per registered tashkeel teacher per month, whereas the Paktika PED spent 4,647 AFN per month per teacher. Average base salary expenditure correlates with the percentage of qualified teachers in a province(r= .65). Some of the provinces with the most qualified teachers (Kabul City, Jawzjan, Nangarhar, Balkh and Kapisa) also spend the most on teacher salaries per month. Conversely, some of the provinces with the least qualified teachers (Uruzgan, Paktika, Kandahar, Badghis, and Zabul) also spent the least on teacher base salaries per month, see Map 9. Map 9: Base salary expenditure per tashkeel teacher (average 2010-2016) and percentage qualified teachers 2016. (Salary per month AFN in color, and qualified teachers in digits) 76% 79% 43% 63% 69% 77% 61% 41% 3,900 - 5,000 AFN 33% 74% 33% 5,001 - 6,000 AFN 18% 60% 47% 79% 6,001 - 7,000 AFN 63% 67% 60% 7,001 - 7,350 AFN 68% 72% 7,351 - 7,700 AFN 40% 66% 32% 12% 7,701 - 8,050 AFN 33% 8,051 – 8,700 AFN 47% 56% 18% 10% 27% 85% KABUL CITY 14% 57% NATIONAL AVERAGE 37% 17% 38% (Source: AFMIS, Payroll Registration Database) 332. There are some noticeable outliers in the 2016 dataset too. Nimroz, Baghlan, Helmand, Herat each had relatively few qualified teachers but spent more than the national average on salaries per month. However, some of these outliers appear to be caused by error and confusion on codes and labels. Nimroz and Herat PED expenditures in 2016 deviated significantly from previous years. Given limited turnover and teacher growth, these fluctuations are likely caused by PEDs relabeling employees or recoding expenditures. 333. It is not possible to assess expenditures per haquzama teacher here, because this study was unable to find accurate and historical data on the number of haquzama contracts. However, using the haquzama remuneration scale discussed in Chapter 0, annual expenditures per qualified haquzama FTE should be between 51,840 AFN and 60,480 AFN (depending on the qualification level of the haquzama teacher, and based on a 9 month per year remuneration). Haquzama teachers cost significantly less because they are only paid for the 9 months they work and not the full calendar year. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 115 6.2.2 Total MoE wages and salaries expenditures per teacher FTE 334. Another method of calculating teacher costs is to divide the total MoE Wages and Salaries expenditures (AFMIS code 21) by the number of MoE employed teachers (in FTEs). This indicator not only reflects average expenditures on teacher base salaries and allowances, but also includes salary and allowance expenditures of support and managerial staff (approximately 19% of all MoE FTEs) per teacher. Given that a rise in students (and teachers) requires additional academic and non-academic staff at schools (e.g. ajeer assistant teachers covering maternity leave), as well as additional staff at PEDs and DEDs, this indicator more accurately predicts total MoE wage bill expenditures per teacher FTE. This indicator will also correct for error created by the incorrect labeling of teachers (in the payroll registration database) and incorrect coding of expenditures (in AFMIS). 335. In 2017, the MoE spent 126,817 AFN (1,812 USD) on total wages and salaries per teacher FTE. This amount not only includes the teacher base salaries and allowances, but also salaries and allowances of support and managerial staff, overtime, as well as other wage expenditures. This amount is calculated by dividing the total MoE wage bill expenditures by the estimated total amount of teacher FTEs. As such, this amount reflects the average wage bill cost of mobilizing one full-time teacher in the field. Note that a key impediment to calculating this indicator is the lack of accurate teacher FTE data (i.e. the number of haquzama teachers). There is no historical or provincially disaggregated data on teacher FTEs, and the total number of teacher FTEs in 2017 could not be determined with full certainty, see Table 10. 6.3 Teacher Costs Forecasting 336. Accurate forecasting of teacher costs is difficult due to several unknown indicators. First, it is unclear why teacher salary expenditures per teacher have increased by 42% between 2010 and 2016. Intra- and inter-bast teacher promotion can only partly explain this increase, and more than half of the 42% increase remains unexplained. While the implementation of the P&G pay scale is a likely contributor to this increase, it is difficult to assess how much this will continue to affect average teacher cost in the next seven years. Second, (intra- and inter-bast) teacher promotion will continue to increase the average cost of a tashkeel teacher, but this trend will likely slow in the coming years (given the limited period in which incremental salary increases occur). However, a proposal to reform the promotion system and allow senior teachers into basts 4 and 3 would further increase teacher costs. Last, and importantly, the current pay scale for civil servants has not been adjusted since its introduction in 2009, while the consumer price index (CPI) rose by 34% (2009-2016). Although it is unclear how and when the government will address this issue, at some point an inflation adjustment of the scale will have to be implemented. 337. In this section, several forecasting scenarios will be presented on teacher costs and MoE wage bill expenditures. The scenarios vary in terms of methods used to determine teacher needs (see, Chapter 0, ‘actual and adjusted class size’ methods), and inflation adjustment. Average teacher costs are adjusted year-on-year for increasing costs due to promotion and P&G reform, as well as inflation. The models all share the following assumptions: I. The relative size of the haquzama teacher workforce (15% to 20%) will remain unchanged, and thus the average wages and salary expenditure per teacher FTE will not be affected by a change in salary composition due to the contract type distribution; II. The percentage of under qualified teachers will continue to decline, and both intra- and inter-bast promotion, as well as finalizing P&G reforms, will continue to increase Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 116 average teachers costs as per the P&G scale. However, the combined impact is expected to be significantly less in coming years, and estimated at a 2.1% year-on-year increase; III. Consumer prices will continue to rise 5% year on year (IMF estimate85) IV. Afghanistan’s nominal GDP will continue to grow 9% year-on-year GDP growth (IMF estimate86) 338. Future teacher costs are determined by average teacher salaries (assumed to increase 2.1% annually due to teacher promotion and limited turnover), the number of teachers, and salary inflation adjustments. Assuming student enrollment trends will continue as in the past seven years, the number of teachers will need to increase by 3.9% annually, to maintain the current PTR (29.8 attending students per teacher FTE), and will need to increase by 5.8% annually to reduce the PTR (25.7 attending students per teacher FTE, as prescribed by the ‘actual class-size’ method). In both cases however, teacher costs will grow slower than anticipated GDP growth (9% y-y), if salaries are not adjusted for inflation. However, in both scenarios teacher costs will exceed anticipated GDP growth, if fully adjusted for inflation, see Figure 41. If education expenditures are to maintain current rates (as compared to GDP), the ‘Maintaining PTR’ scenario allows for a 3% annual inflation adjustment, whereas the ‘Reducing PTR’ scenario allows for only a 1.1% annual inflation adjustment. Figure 41: Teacher wage bill growth compared to GDP in percentages 5.0% 09% 5.0% 5.8% 9.0% 3.9% 2.1% 2.1% 00% Maintaining PTR Reducing PTR GDP Growth (estimate) Teacher costs increase (estimate) Teacher number increase Inflation Adjustment Source: Staff Calculations 85 World Economic Outlook (April 2018). International Monetary Fund. 2018. 86 Ibid. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 117 If teacher (and other MoE) salaries are not adjusted for inflation, the MoE wage bill will increase by 59% in 2025, totaling 48.2 billion AFN, to maintain the current PTR. Reducing the PTR will increase the MoE wage bill by 84% in 2025, totaling 55.6 billion AFN, see Figure 42. Figure 42: Projected MoE wage bill expenditures 2018-2025 (in billion AFN), without inflation adjustment 55.6 51.5 47.7 44.2 41.0 48.2 38.0 45.5 35.2 42.9 32.6 40.5 30.3 38.2 34.0 36.0 30.3 32.1 1396 1397 1398 1399 1400 1401 1402 1403 1404 2017 2018 2019 2020 2021 2022 2023 2024 2025 Maintaining PTR (29.8), no inflation adjustment Reducing PTR (25.7) no inflation adjustment Source: AFMIS, Staff Calculations If teacher (and other MoE) salaries are fully adjusted for inflation (5%), the MoE wage bill will increase by 130% in 2025, totaling 69.7 billion AFN, to maintain the current PTR. Reducing the PTR will increase the MoE wage bill by 164% in 2025, totaling 79.9 billion AFN, see Figure 43. Figure 43: Projected MoE wage bill expenditures 2018-2025 (in billion AFN) with full inflation adjustment 79.9 70.7 62.6 69.7 55.5 62.8 49.1 43.5 56.6 38.6 51.0 34.2 45.9 30.3 41.4 37.3 33.6 30.3 1396 1397 1398 1399 1400 1401 1402 1403 1404 2017 2018 2019 2020 2021 2022 2023 2024 2025 Maintaining PTR (29.8), with inflation adjustment Reducing PTR (25.7) with inflation adjustment Source: AFMIS, Staff Calculations Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 118 339. Maintaining current PTRs (3.8% y-y teacher growth), combined with anticipated average salary increases due to promotion (2.1% annually) and partial inflation adjustments (3% annually) will not significantly alter education expenditures compared to GDP (assuming 9% GDP growth). Full inflation adjustment of teacher salaries, and maintaining the current PTR, would result in a progressive increase of education wage bill expenditures (compared to GDP), resulting in similar wage bill expenditures in 2025 as in 2011. Only a decrease in current PTRs (increasing the number of teachers faster than anticipated student growth) and full inflation adjustment of MoE salaries, would significantly increase the MoE wage bill by 2025, resulting in an estimated 2.75% of nominal GDP, see Figure 44. Figure 44: Actual and projected MoE wage bill expenditures as percentage of GDP (2011-2025) 0.029 2.75% 0.027 0.025 2.44% 2.40% 2.39% 0.023 2.34% 2.33% 2.24% 0.021 2.06% 2.00% 0.019 1.92% 1.86% 0.017 1.66% 0.015 1390 1392 1394 1396 1398 1400 1402 1404 2011 2013 2015 2017 2019 2021 2023 2025 Maintaining PTR (29.8), no inflation adjustment Reducing PTR (25.7) no inflation adjustment Maintaining PTR (29.8), full inflation adjustment Reducing PTR (25.7) full inflation adjustment Actual (Source: AFMIS, IMF, and staff calculations) 340. Furthermore, total government expenditures are expected to increase faster than GDP growth87, and a fixed commitment to education expenditures (i.e. as a percentage of total government expenditures) would allow for additional fiscal space to reduce PTRs (i.e. applying the ‘actual class size’ method) or increase other education expenditures (O&M and development investments). 87 World Economic Outlook (April 2018). International Monetary Fund. 2018. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 119 6.4 Discussion and Recommendations 341. Teacher base salaries alone are responsible for half of all MoE expenditures, and the combined wage bill, including teacher allowances, overtime, and non-teacher wages, absorbs 78% of the MoE budget. Transparent management of the MoE wage bill, with deliberate planning and strong oversight, is key to promoting efficiency and equity in the education sector. This chapter compared teacher wages and salaries to other education expenditures and calculated average teacher costs. It also analyzed per student and teacher expenditures and concluded by presenting different forecasting scenarios on the MoE wage bill. 342. Maintaining the status quo (maintaining pay levels and pupil-teacher ratios) will require the MoE’s budget to increase by around 11% year-on-year (assuming inflation and enrollment trends will continue). This 11% includes the rising number of teachers needed in front of the class in Afghanistan, the rising average costs of a teacher, and the rising prices of consumer items, for which teacher salaries will need to be adjusted. Since 2014, the MoE budget has plateaued and per student expenditures decreased somewhat, despite the rising costs of teachers. Education expenditures as percentage of total government expenditures or as percentage of GDP have decreased significantly. A continued depletion of per student expenditures will cannibalize the MoE’s budget and result in larger classes and a decrease in teacher’s real income. 343. Developing a more predictable education budget process combined with a more accurate and comprehensive wage bill overview will enable the MoE to make better and more transparent decisions on priorities for teacher workforce growth and salary adjustments. 344. Education services are not provided equally across Afghanistan’s provinces. The net pupil-teacher ratio (number of attending students per teacher FTE) in Panjshir is estimated at 14, whereas in Helmand this is 42, and in Khost this is 38. Wage and salary expenditures per student in Panjshir are also almost three times higher than in Khost and more than twice that in Helmand. In Panjshir, eight out of ten teachers is qualified, whereas in Helmand less than four out of ten teachers is qualified, and even less teachers are subject qualified. 345. The combination of bottom-up planning, ambiguous planning standards and input- based budgeting reinforces provincial inequalities. A norm-based budgeting approach, allocating resources based on the number of attending students, would significantly increase resources in provinces like Khost, Helmand, Daikundi, Uruzgan, and Badghis. This would enable these provinces to not only address teacher shortages but also promote retention of qualified teachers and incentivize teacher training and recruitment. Existing planning and budgeting procedures perpetuate the relative deprivation of these provinces, and reforms are required to enable schools in these provinces to catch up on the progress made elsewhere in Afghanistan. 346. Last, the analyses of AFMIS ‘overtime’ expenditure data per province shows again the lack of regulation or a clear policy on the allocation of temporary contract teachers. Haquzama teachers are not exclusively working in those provinces where they are most needed and year- on-year, there are dramatic fluctuations in the number of temporary teachers per province. This confirms the need for clear guidelines and planning standards for temporary teachers as well as the need for strict oversight and accountability mechanisms. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 120 Recommendations – Teacher Costs The Ministry of Education should (jointly with the IARCSC and the Ministry of Finance): SHORT TERM: (within the next 12 months) 347. Develop a more predictable education budget process (within the medium-term fiscal framework) to enable planning for the MoE, as well as PEDs, DEDs and schools. Reviewing expenditures relative to GDP would enable the MoE to more sustainably plan the wage bill. PEDs with indicative multiyear budget commitments will be empowered to develop multi-year plans to address shortages in pre-service trained applicants and improve the teacher GPI. 348. Develop a policy note that transparently identifies and illustrates the MoE’s fiscal requirements, and the effects of limited budget growth over the past years. The policy note should mobilize consensus across government and development partners for sustained growth of the education budget and advocate for a minimum percentage growths based on student enrollment, costs and inflation adjustments. MEDIUM TERM: (within the next 36 months) 349. Work progressively towards norm-based budgeting with per student allocations to provinces, districts and schools. This will require the delegation of more budget authorities to the provincial education directorates. Provinces with relatively low per student expenditures currently, will progressively receive more budget. PEDs need more flexibility to decide how excess funds can be spent, to stimulate progress in teacher shortages, quality, training and gender. A clear policy, regulations, and strong oversight are required to identify eligible expenditures for these provinces, to promote efficiency and prevent abuse. Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 121 7 References Altai Consulting. (2015). Education Sub-National Assessment: Strengthening Public Education Management. Independent Joint Anti-Corruption Monitoring and Evaluation Committee. (2015). Vulnerability to Corruption of Teacher Recruitment in the Ministry of Education. Independent Joint Anti-Corruption Monitoring and Evaluation Committee. (2017). Ministry-Wide Vulnerability Assessment to Corruption Assessment of the Ministry of Education. Ministry of Education, Islamic Republic of Afghanistan. (2016). National Education Strategic Plan: 2017-2021. Ministry of Education, Islamic Republic of Afghanistan: Teacher Education Directorate. (2016). Teacher’s Mapping & Training Projection Survey Report. The World Bank. (2018). The Learning Crisis in Afghanistan, Results of the SABER Service Delivery Survey 2017 (PRELIMINARY DRAFT). UNICEF. (2017). Afghanistan Annual Report 2017. The Impact of Number of Students per Teacher on Student Achievement (2015). N. Cok & C. Belik. Procedia: Social And Behavioral Sciences 177 (2015) p65-70. Photo 5: Boys class (Kabul City 2018) Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 122 8 Appendices & Annexes 8.1 Appendices Appendix 1: Overview of MoE contract types TASHKEEL "BILMAKTA" (NON-TASHKEEL) "Haquzama inside MoE Admin / Teachers Ajeer Haquzama outside Ajeer / Bilmakta tashkeel" Management Staff # BASED ON PAYROLL Unknown 192.560 18.901 43.065 15.877 2.531 (EARLY)2018 # BASED ON DATA PAYROLL Unknown 203.641 18.661 35.033 13.690 1.690 (LATE)2017 AFMIS 2016 EXP - Unknown Unknown 261.3 28.3 34.3 4.7 in million USD (part of 52.9) (part of 52.9) Haquzama is This contains the Haquzama is unique HR bulk of teacher This is a low skilled unique HR This contract type modality for contracts, but also labor / support modality for Bilmakta is a contains all Academia - with its the contracts of staff contract - Academia - with its government-wide ADMIN/MANAGE own regulation - other academic typically guards, own regulation - temporary contract Description of MENT staff Central INSIDE Haquzama staff such as cooks, drivers, tea OUTSIDE modality to hire contract pool Ministry and PED are 'fixed-hour" principals, makers, cleaners - Haquzama are both unskilled and and DED (but also ADDITIONAL headmasters, lab- but also temporary - full- skilled (mid-level) all academic contracts for and librarians. uneducated time - teacher workers supervisors???) existing academic Academic teachers contracts for non- POOL INCLUDES tashkeel staff Supervisors?? tashkeel SOME - under SOME - under MOST - unknown MOST - unknown qualified (<12G) qualified (<12G) ALL - most or all how much but how much but NONE - NO can be labeled as can be labeled as Includes teachers? individuals in this expected >95% of expected >95% of TEACHERS Ajeer Tashkeel, Ajeer Bilmakta, group are teachers total total unclear how much unclear how much % of this group % of this group Temporary Temporary "Overtime/additio Permanent civil Permanent civil Permanent civil Type of contract government government CONTRACT nal work" contract servant servant servant contract contract Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 124 usually 9 months duration of (but also 3 months Permanent Permanent Permanent 9 months 9 months contract - covering for maternity leave) 5 - 6 (typically) but 3 - 6 (typically) but also higher grades, also 1 and 2 for Grade ? 7 and 8 (always) no grade no grade for principals and very senior school directors positions As per Academic As per civil service As per civil service Promotion N/A Promotion N/A N/A law law regulation Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 125 In exceptional In exceptional In exceptional cases the contract cases the contract cases the contract holder will holder will holder will terminate his terminate his terminate his contract before the contract before the contract before the end of the 9 end of the 9 end of the 9 Contract duration Permanent Permanent Permanent months - in which months - in which months - in which case a new short- case a new short- case a new short- term contract can term contract can term contract can be prepared to be prepared to be prepared to finalize the initial 9 finalize the initial 9 finalize the initial 9 months months months Work months/year 9 months / year 9 months / year 12 months / year 12 months / year 9 months / year 9 months / year Fixed hours depending on academic position: 40 hrs. / Week - 40 hrs. / Week - WORKLOAD 24 - 28 hrs. / Week teachers +100% minus 1hr lunch minus 1hr lunch - typical teaching 24 - 28 hrs. / Week Hours per week (=full-time) but per day - small per day - small hours per week - - Fixed hour full- are fixed - but Work hours/week other academic variation between variation between there is no non- time teacher negotiated in the staff Kabul based and Kabul based and teaching hrs. contract contract. (principals/libraria non-Kabul based non-Kabul based obligations ns) gets +25% per employees employees additional school shift Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 126 0 leave days - 0 leave days - 0 leave days - absent days will be absent days will be absent days will be Leave days 30 days / year 50 days / year 50 days / year deducted from deducted from deducted from salary salary salary Salaries 9 months / year 12 months / year 12 months / year 12 months / year 9 months / year 9 months / year Determined by Determined by Salary is negotiated Haquzama Bill: 60, 6.500 AFN (G6.S1) Haquzama Bill: 60, - without strict 70, or 80 AFN per till 16.000 AFN 70, or 80 AFN per guidelines - from hour - depending 6.500 AFN (G6.S1) (G3.S5) - but 5.000 AFN (G8.S1) hour - depending 5.000 AFN (min) to Salary on level of till 9.200 AFN exceptionally also to 6.200 AFN on level of 13.000 AFN (max) education (G5.S5) 18.400 AFN (G2.S1) (G7.S5) education SALARY - note this is the (MAX=96 hrs per to 32.500 AFN (MAX=96 hrs per salary range for month) 5.760 AFN (G1.S5) month) 5.760 AFN G8.S1 to G4.S5 - 7.680 AFN - 7.680 AFN SALARY AFMIS Teachers (Primary, Civilian Employee Civilian Employee OVERTIME OVERTIME Contractor LABEL Secondary, Others) (Permanent) (Worker) SALARY AFMIS 21124 21107-08-09 21100 21101 21124 21110 CODE Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 127 PAID OVERTIME NO NO YES - paid per hour NO NO NO AFMIS CODE 21124 OVERTIME YES - GR12=8AFN; YES - GR12=8AFN; GR14=12AFN; GR14=12AFN; Academic top-up / UNDERGR=2.000AF UNDERGR=2.000AF NO NO NO NO month N; N; MASTER=4.000AFN MASTER=4.000AFN ; PHD=6.000AFN ; PHD=6.000AFN 21121 Professional 21121 Professional AFMIS CODE N/A NO NO NO Cadre Allowances Cadre Allowances Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 128 Appendix 2: Overview of cadre allowances by program and academic title Cadre Allowances by Program No Teaching Position/Title (Academic) Academic Cadre Allowance TTCs Cadre Allowance Scale 1 Showondae 12,000 2 Showonmal 10,500 3 Showanyar 9,000 4 Shohyalae 8,000 TVET Cadre Allowance Scale 1 Maslak Shanas 12,000 2 Maslak Fahm 10,500 3 Maslak Amoz 9,000 4 Mawen Maslak Amoz 8,000 Science Center Cadre Allowance Scale 1 Shohand 12,000 2 Showonwal 10,500 3 Showondae 9,000 4 Showonmal 8,000 5 Showonyar 7,000 6 Shohyalae 6,500 Editing and Translation Cadre Allowance Scale 1 Sarmowalef 12,000 2 Mawen Sarmowalef 10,500 3 Mowalef 9,000 4 Mawen Mowalef 8,000 Islamic Education Cadre Allowance Scale 1 Sar Modares 12,000 2 Mawen Sar Modares 10,500 3 Modares 9,000 4 Mawen Modares 8,000 8.2 Annexes Annexes are attached in separate files to this report. Annex 1: Educational Institution’s Planning Standards Annex 2: Two Terms of References developed for new teacher positions Annex 3: Application Form Annex 4: Civil Servant’s Law Annex 5: Civil Servant rules and Regulations Annex 6: Pay and Grades Scale Annex 7: Bilmakta bill Annex 8: Haquzama bill Teacher Policy Study: Towards a more Equitable and Efficient Use of Education Resources in Afghanistan 130