WORLD BANK GROUP Enhancing Human Capital by Improving Education November 2021 Kosovo Country Economic Memorandum  1 Kosovo Country Economic Memorandum Enhancing Human Capital by Improving Education Table of Contents Overview 11 Improving Education Quality 23 School Optimization Scenario for Pre-University Education 29 Simulation of network optimization in the municipality of Suharekë 35 Skills Gap Discussion 39 Impact of COVID-19 on the Education System 45 Summary and Policy Recommendations  51 References 66 2  Figures Fig 1. Gross Enrollment for all Levels, Fig 10. Spending on Pre-university Fig 19. Mean Literacy Scores, Total Percent, 2014–19 p.13 Education by Category, Working Age Population Fig 2. Lower and Upper Secondary 2009–19, Euros p.19 and Employed (aged 15-64), Enrollment in Kosovo and Fig 11. Kosovar Salaries in the Western Balkans, FSU, and Comparator Countries, Percent Education Sector and Other OECD Countries p.40 p.13 Comparable Sectors, 2018 p.20 Fig 20. Distribution of Core Literacy Fig 3. Population of School-age Fig 12. Per-student Costs for Pre- Scores by Demographics, Children Projected through university Education by Percent, 2018 p.40 2061, Thousands p.14 Municipality p.21 Fig 21. Distribution of Core Literacy Fig 4. Children Aged 36–59 Months Fig 13. Student-Teacher Ratios by Score by Education and Labor Developmentally on Track in Municipality p.21 Market Status, Percent, 2018 Literacy-Numeracy, Percent p.41 Fig 14. Student-Teacher Ratios in p.16 Kosovo, 2014–19 p.24 Fig 22. Worker’s Perceptions of How Fig 5. Fourth Grade Proficiency in Their Education and Skills Fig 15. Percent of Students whose Math and Science, Kosovo and Match Their Jobs p.42 School Principals Responded Comparators, Percent p.16 “To some extent” or “A lot” to Fig 23. Education and Jobs Matching, Fig 6. 15-year-old Students the Question: Is your school’s Kosovo and Comparators p.43 Performing Below Functional capacity to provide instruction Fig 24. Estimated Impact of COVID-19 Literacy in Reading, PISA 2018, hindered by any of the on PISA Scores p.46 Percent p.17 following issues? p.26 Fig 25. Estimated Impact of COVID-19 Fig 7. Matura 12 Examination Results Fig 16. Distribution of the Types of on the Socio-economic by Municipality, 2019 p.17 School Units in Kosovo, 2016 Achievement Gap p.47 Fig 8. Kosovo’s Spending on p.31 Fig 26. School Resources for Resilient Education Compared, Percent Fig 17. Principle of the School Education Compared, Percent of Total Government Spending Network Optimization p.48 and Percent of GDP p.18 Algorithm p.32 Fig 9. Education Spending, Kosovo, Fig 18. Map of School Units in 2007–19, Euros p.18 Suharekë p.36 Tables Annexes Table 1.  School Facilities in Kosovo Annex 1: Mapping of Gross Enrollment With Inadequate Infrastructure by Type By Level of Education p.55 of Problem, 2016 p.25 Annex 2: Details of the Algorithm for Table 2.  School Facilities in Suharekë, School Network Optimization and by Type of Infrastructure Problem, 2016 Identification of Factors That May p.37 Impact A Simulation p.63 Table 3.  Estimated Savings for Suharekë in Three School Network Optimization Scenarios p.37  3 © 2021 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. 4  Acknowledgements This note was prepared by Angela Demas (Senior Education Specialist), Syedah Aroob Iqbal (Research Analyst), and Paul Cahu (Consultant) with contributions from Mrike Aliu (Human Development Specialist) and Besart Myderrizi (Research Analyst) as part of the Kosovo Country Economic Memorandum, which was led by Asli Şenkal (Senior Economist, Task Team Leader). This work was overseen by Linda Van Gelder (Country Director, Western Balkans), Massimiliano Paolucci (Country Manager, Kosovo and North Macedonia), and Harry Patrinos (Practice Manager, Education Global Practice). The analysis benefitted from comments from Enrique Blanco Armas (Lead Economist, Western Balkans) and Harry Patrinos. The team is grateful to the Ministry of Education, Science, Technology, and Information (MESTI) for their support in data gathering and comments, in particular to Agim Berdyna (former General Secretary), Xhavit Rexhaj (former Deputy Minister of Education and Science), Fatmir Elezi (Head, Division for Standards, Monitoring, and Evaluation), Enver Mekolli (Management Information Systems Officer), Rina Ponosheci (Head, Infrastructure), and Construction Management and Consulting (CMC). In addition, the team thanks the MESTI working group for school network optimization for providing valuable feedback and discussion on the school network optimization simulation.  5 Acronyms AI Administrative Instruction ECD Early childhood development GDP Gross domestic product KESP Kosovo Education Strategic Plan MESTI Ministry of Education, Science, Technology and Innovation MICS Multiple Indicators Cluster Survey OECD Organisation for Economic Co-operation and Development PISA Programme for International Student Assessment STEP Skills Towards Employability and Productivity TIMSS Trends in International Mathematics and Science Study UNICEF United Nations International Children’s Emergency Fund 6  Executive Summary This note assesses pre-university education in Kosovo and identifies policies that can improve education quality and equity. There is a substantial need for improvement to the foundations of the education system, especially in terms of coverage and quality. If Kosovo is to reap the benefits of its demographic dividend for faster growth and poverty reduction, making it a policy to invest in human capital through education must be a priority. To tap into the potential of human capital to enhance economic growth, Kosovo must therefore invest in the current generation of children and youth to ensure high human capital and labor productivity in the coming decades. In 2020, Kosovar children of age 17 and younger constituted 30 percent of the population (KAS). This is much higher than the Europe and Central Asia average of 18 percent and the regional average, when high-income countries are excluded, of 21.4 percent.1 However, as this percentage has been gradually declining, now is the time for action. The low levels of foundational literacy skills limit the ability of Kosovars to develop the skills the labor market requires,  especially for those living in rural areas and minority communities and those of low socioeconomic status. As evidenced by national and international benchmarks2, education in Kosovo does not deliver the skills needed for a productive life. Low access to early childhood education and poor preschool coverage negatively impacts development of school readiness as measured by foundational literacy and numeracy. In 2019, only 25 percent of children age 36–59 months were developmentally on track for literacy and numeracy (UNICEF 2020). In the years that follow, Kosovar students do not catch up and remain well-below functional literacy and numeracy. By age 15, 79 percent are functionally illiterate (OECD 2018). Even though most students in Kosovo complete 13.2 years of schooling by age 18, when that number is adjusted for quality of learning, it is equivalent to just 7.7 years of school (HCI 2020).  As a result, adults have low literacy skills. In the 2018 STEP (Skills Towards Employability and Productivity) Household Survey, nearly 14 percent of respondents3 failed the core literacy test.  Kosovo has made considerable progress in increasing access to preprimary and primary education, but more needs to be done to equitably increase coverage of preschool and secondary education. Only 16 percent of 3–5-year-olds have access to any kind of preschool education, and about 30 percent of municipalities have no public preschool institutions. In secondary education Kosovo is struggling to maintain enrollment: in lower secondary, between 2014 and 2019 gross enrollment dropped from universal to 91.5 percent. In 2019 in upper secondary gross enrollment was 83.4 percent, lower than in neighboring countries.  Equitably increasing access to quality preschools by promoting a variety of types of preschool providers gets to the heart of establishing school readiness and acquisition of foundational literacy and numeracy skills. To ensure better completion rates in secondary, tracking student performance and setting up a warning system for early school leavers can help in targeting compensatory services to vulnerable students. For example, tutoring works well in recovering learning losses. A 3-month tutoring program can help restore the loss of up to 5 months of normal schooling,4 and can improve math and language performance by almost 5 percent.5 Narrowing the wide geographic, ethnic, and socioeconomic disparities in education outcomes and ensuring that all children have access to quality education is paramount.  Foundational skills are low for all children, but the learning divide widens for rural and minority students as they grow. Among children living in Roma, Ashkali, and Egyptian communities, about 18 percent have adequate reading skills and about 13 percent have similar math skills.6 Moreover, for children living in Roma, Ashkali and Egyptian communities, attendance is much lower, at primary school averaging 84 percent versus 100 percent for the general population,  at lower secondary 63 versus 93 percent, and at upper secondary 31 versus 87 percent (UNICEF 2020). 1 World Development Indicators (WDI) 2 National Matura 12 Examination, Human Capital Index (HCI), Learning Poverty, TIMSS, and PISA results, etc. 3 In all countries, the target population for the World Bank’s STEP Skills Measurement Program is urban adults aged 15 to 64, whether employed or not. 4 The Economist. 2020. “England’s catch-up tutoring programme has bold ambitions.” https://www.economist.com/britain/2020/11/05/englands-catch-up-tutoring-programme-has-bold-ambitions 5 EurekAlert. 2021. Op. Cit. 6 Foundational reading skills are measured as the percentage of children aged 7–14 who can: (1) read 90% of words in a story correctly, (2) answer three literal comprehension questions, and (3) answer two inferential comprehension questions. Foundational math skills are measured as percentage of children aged 7–14 who can successfully perform: (1) a number reading task, (2) a number discrimination task, (3) an addition task and (4) a pattern recognition and completion task.  7 Income inequalities also exacerbate the education gap: students in the poorest quintile fall behind by roughly one year of schooling attainment to the richest quintile. There are deficits in the quality of education even though Kosovo spends about as much on education as other middle-income countries. Kosovo, in 2019, spent 4.6 percent of GDP and 16 percent of total government spending on education, similar to other middle-income countries in the world, but schools in the country face deteriorated or inadequate school infrastructure and severe shortages of educational resources critical to the teaching and learning process. Spending on pre-university education more than doubled over the last decade, but it was mainly driven by an increase in public wages, which account for about 89 percent of overall education spending. Demographic changes (declining population, migration) offer an opportunity for savings in the education budget that can be directed to quality enhancement. Municipalities vary considerably in terms of both spending and outcomes. A more equalized spending-per-student approach might lead to savings required to increase spending on quality-enhancing inputs and educational resources for schools. There is also potential for creating fiscal space for quality investments by optimizing school networks. To illustrate the scope for possible savings, this note conducts an exercise of school network optimization for one medium-sized municipality in Kosovo. To effectively lead the process, a detailed update will be required of the school, student, and teacher mapping by education institution/facility,7 a cooperative collaboration between the Ministry of Education, Science, Technology and Innovation (MESTI) and municipalities, and buy-in from the local community. Quality of education in Kosovo is hindered by lack of adequate infrastructure and educational materials.  Based on a national school mapping exercise supported by the World Bank in 2016, 80 percent of schools lacked drinkable water or proper sanitation and 70 percent lacked laboratories. Similarly, a large proportion lacked facilities for sports and recreational infrastructure and IT equipment. Post-COVID, the dearth of adequate sanitation in school facilities is particularly concerning. There’s a strong need to invest in quality-enhancing school inputs to support teachers and schools to deliver quality education. Although there is no shortage of teachers, school principals for 86 percent of 15-year-old students note that instruction is hindered by lack of such educational materials as textbooks, IT equipment, library and laboratory material (OECD 2018). The COVID-19 pandemic is expected to further erode Kosovo’s modest improvements made in education. COVID-19 forced Kosovo to close schools from March through June 2020, which is likely to have lowered learning effectiveness given unequal access to and the poorer quality of education through online or TV platforms in comparison to in-class instruction. Assuming a student gains 40 PISA points of learning in a year, it is estimated that learning in Kosovo will drop by the equivalent of 9 PISA points in reading. The learning losses would translate into an increase in students performing below functional literacy by a percentage point, from 79 to 80 percent. The differences in access to online learning and effectiveness of remote teaching are likely to exacerbate regional and socio-economic gaps. Giving priority to efforts to safeguard education outcomes is therefore important.8 Employers in Kosovo note the poor quality of the education system and the limited practical and professional skills students acquire. Three out of 4 firms that attempted to fill a higher skills position and 3 out of 5 firms that tried to fill a medium to lower skills position encountered problems because of lack of either skills or experience (World Bank 2018a). Furthermore, nearly 60 percent of employers interviewed thought that general education did not provide students with practical experience, and almost half of them noted that it also does not provide up-to-date knowledge or such socio-emotional skills as discipline, timeliness, or people skills. Although employers are more satisfied with TVET outcomes, 35–42 percent think that training systems do not provide people with the necessary skills.  7 School Map Report National Level, 2016. 8 World Bank. 2021. Learning Recovery after COVID-19 in Europe and Central Asia: Policy and Practice. World Bank, Washington, DC. 8  Recommendations: Increasing access and maintaining enrollments 1. Increasing access to preschool education requires expansion of alternative financing models for preschool (beyond public provision), providing incentives for private providers and hybrid public/private models, as well as ensuring minimum quality standards through licensing, monitoring, and training.9 2. At lower and upper secondary, it will be important to institutionalize or continue supporting remedial and extra support programs delivered now through NGOs (e.g. learning centers) targeted especially to low income families, minority children, and rural students. Additionally, the student registry should be developed to better track academic performance over the student’s school career. This registry should also allow for an early warning system of at risk/low performers for better targeting of services that keep students in school. 3. Improving the targeting and coverage of the social assistance scheme to cover poor families with several children10 would also support increasing access and maintaining enrollment through reducing the liquidity constraints. Quality and Equity 4. Continue implementing and improving a system of learning assessments that guides teacher instruction and motivates improvement over time, especially in the lagging municipalities.11 Kosovo would benefit by building on the strengths of a qualified and motivated teaching force, implementing a strong professional development program that caters to teachers’ needs and improving quality of teaching and learning in the country. 5. Provide more financial resources and utilize savings from more efficient spending to supply schools with adequate and good quality educational materials and infrastructure for a safe and conducive learning environment. 6. Introduce a comprehensive plan for digitizing education delivery and ensuring equity of connectivity, sufficient equipment, and skilled/supported teachers who are prepared to use technology to assist learning, especially for children from rural, minority communities, and low socio-economic status. Efficient spending (to enable better quality) 7. Take action to improve the school financing formula based on the recent study supported by the KESIP project offering solutions for the current formula’s implementation challenges and options to introduce equity and efficiency into the formula. Adequate implementation of the revised formula requires providing capacity building at the municipal level. 8. Improve the monitoring and evaluation systems to track results of the dynamic changes in the school funding formula, to make decisions, and to hold stakeholders accountable to achieve better learning outcomes. 9. Conduct a careful study of the distribution of administrators, teachers and specialists could go hand in hand with studies at the municipal level on school network optimization. Retraining opportunities for interested teachers where there is oversupply of teachers for subject areas or levels may help to redirect them to areas where there is a deficit of supply. The strongest teachers should also be encouraged to teach at the early years where foundational skills are being taught. As a first step Ministry of Education should update the school mapping report produced in 2016 and the corresponding education management information system (EMIS). At the same time work with interested municipalities actively seeking to realize savings that can be invested in school quality. 9 For a complete set of recommendations for ECD expansion, please see: Demas, Aliu, Coll-Black, Zafeirakou, Hankey, and Gotcheva. 2021. A Situational Analysis of Early Childhood Development Services in Kosovo. World Bank, Washington, DC. 10 Per current legislation, families may lose their status as social assistance beneficiary if a child turns 18 and has completed compulsory education but is still in full-time higher education (post-secondary TVET or university) (for those in program Category 1), or if the youngest child in the family turns five years of age (for those in program Category 2). The concept note for the new Law on SAS would eliminate this and other discriminatory categorical filters. 11 In accordance with the National Framework for Student Assessment.  9 01 Overview Kosovo has made considerable improvements in extending access to education and has achieved almost universal access for primary education. Primary education in Kosovo starts at age 6 and lasts five years. The gross enrollment rate in primary has consistently increased over time and is now almost universal (Figure 1). Similarly, with increased national focus on pre-primary education, Kosovo has significantly expanded access to pre-primary (grade 0) and is aiming to increase coverage for preschool which falls far behind. Education in Kosovo is organized in six levels, starting with early childhood education and care (ECEC) for children aged 9 months to 6 years, consisting of nurseries for children aged 9 months up to 3 years and pre-school for children 3<5 years,12 according to Administrative Instruction (AI) 19/2016 (MESTI) and preprimary for children aged 5 to 6.13 However, participation in preschool, remains low due to lack of investment, lack of infrastructure, and lack of awareness among households and policy makers about the importance of preschool education (EU 2019). The Kosovo Education Strategic Plan (KESP) 2017–2021 brings preschool and preprimary education to the fore and sets specific targets for both. While preprimary gross enrollment rates have improved substantially over time, from 78 percent in 2014 to 91 percent in 2019, time spent per week in pre- primary averages only 10–12 hours. Preschool enrollment rates are still very low (Figure 1). In 2019 gross enrollment in preschool (for children aged 3 to <5) was 11 percent, much lower than the KESP target of 20 percent. Preschool education in Kosovo is still very sparsely available, and 11 municipalities14 have no public pre-school institutions or kindergartens. Scarcity of preschools and lack of investment in preschool education has led to limited preschool enrollment across Kosovo. In 2019, there were only 44 public preschools/kindergartens in the entire country (MESTI and KAS 2019/2020). However, the private sector contributes significantly to provision of preschool services through private and community-based kindergartens that account for 55.7 percent of the total preschool (0-5 ages) enrollment in the country.15 High quality early childhood development programs can deliver high returns per child on upfront costs through better outcomes in education, health, employment and social behavior in the decades that follow.16 Failing to invest in services to support ECD can lead to long term costs, not only for children, especially those most disadvantaged, but also for their mothers, families, and for the human capital of the country. Recent research reveals returns on investments to high- quality early childhood programming to be even higher than previously estimated, delivering a 13 percent annual return on investment.17 Children who participate in high quality early childhood programming have stronger outcomes both short- and long-term, in education, health, and in economic productivity. Parents also benefit, with improved adult labor outcomes found for both child participants and their parents. Kosovo is struggling to expand access for lower and upper secondary education. Lower secondary in Kosovo lasts four years for children aged 11–14 and upper secondary lasts three years for children aged 15–17, including gymnasium and vocational education. Gross enrollment in lower secondary slipped from universal in 2014 to 91.5 percent in 2020. Gross enrollment in upper secondary went up slightly from 2014 but in the last two years has been trending downward at 83.4 percent in 2019. Kosovo’s enrollment in lower and upper secondary are lower than the 2018 figures from neighboring countries (Figure 2): Albania (98% lower secondary and 92% upper secondary), Montenegro (92% and 88%), and Serbia (101% and 90%). Also, the comparatively lower enrollment rates and the declining trend are despite the gradual continual decline of the school-age population in Kosovo (Figure 3). 12 Some pre-schools also include children aged 5-6, but the majority of children transition to pre-primary classes at this age. 13 The first law on Preschool Education and the first curriculum for early childhood education (0 – 6 years old) were passed in 2006. 14 These are: Decan, Dragash, Hani I Elezit, Junik, Leposaviq, Navoberde, Prizren, Shterpce, Zubin Potok, Zvecan and Kllokot. 15 Author’s calculations based on data from Report: Education Statistics in Kosovo 2019/20. 16 Retrieved from: https://heckmanequation.org/resource/the-heckman-equation-brochure 17 Ibid 12 Overview Figure 1. Gross Enrollment for all Levels, Percent, 2014–1918 100% Pre-school (3 - 5 year old) Pre-primary (5 - 6 year old) 80% Primary (Grades 1 to 5) Lower Secondary (Grades 6 to 9) 60% Upper Secondary 40% 20% 0% Source: (1) KAS data; World Bank staff calculations; (2) KAS 2019 and UNESCO Institute of Statistics 2018. Note: goes with Figure 1: Enrollment covers both public and private schools. Data do not include population and enrollment numbers from the following Serbian-majority municipalities: Mitrovice, Ranillug, Gracanice and Pratesh. Enrollment numbers are incomplete for other Serbian-majority municipalities: Zubin Potok, Leposaviq, Zvecan, Sterpce, Kllokot and Novoberdo. Figure 2. Lower and Upper Secondary Enrollment in Kosovo and Comparator Countries, Percent Gross Enrollment Rate 120 98 101 101 100 91 92 92 90 87 88 88 81 80 68 60 40 20 0 Republic of Kosovo Armenia Montenegro Albania Serbia Kyrgyzstan Moldova Lower secondary Upper secondary Source: (1) KAS data; World Bank staff calculations; (2) KAS 2019 and UNESCO Institute of Statistics 2018. Note: goes with Figure 1: Enrollment covers both public and private schools. Data do not include population and enrollment numbers from the following Serbian-majority municipalities: Mitrovice, Ranillug, Gracanice and Pratesh. Enrollment numbers are incomplete for other Serbian-majority municipalities: Zubin Potok, Leposaviq, Zvecan, Sterpce, Kllokot and Novoberdo. 18 Preprimary enrollment rates include 5-year-old children enrolled in kindergartens. Overview 13 Figure 3. Population of School-age Children Projected through 2061, Thousands 500 Population Thousands 400 Upper Secondary (15 - 17 year olds) 300 Lower Secondary (11 - 14 year olds) 200 Primary (6 - 10 year olds) 100 5 year old 3 & 4 year olds 0 Source: KAS. There are also wide municipal disparities in access to education in the country. Though nationally the primary gross enrollment is close to universal, enrollment in certain municipalities is quite low, among them Junik (60%) Decan and Dragash (67%), and Kamenicë (75%).19 A few municipalities like Fushë Kosova, Obliq, and Vushtrri have much higher gross primary enrollment rates than 100, which suggests that a number of children that are not of primary age might be enrolled in primary education. However, ages of students in different classes will be needed to understand the high gross enrollment rates in these municipalities. Gross enrollment rates for preprimary, lower, and upper secondary show similar variation across regions with Decan, Dragash, and other municipalities falling behind (see Annex 1 for gross enrollment rates mapped across municipalities by level of education). Poor families, especially those in rural areas having several children may also be at a disadvantage accessing education services. While 18 percent of the country’s population lives below the poverty line, estimated at 1.85 euros per day, poverty is higher in rural areas at 19.4 percent of the total rural population. For rural children ages 0-18, it increases to 24 percent. Among all households, families with more children are more likely to be poor, with the poverty rate of households with three or more children being 2.5 times that of households without children. Children from Roma, Ashkali, and Egyptian communities also face higher risk of both poverty and severe poverty when compared to the population overall. Improving the targeting and coverage of the social assistance scheme to cover poor families with several children would also support increasing access and maintaining enrollment through reducing cash constraints (Box 1). 19 Authors’ calculations using KAS estimated populations of primary school age and enrollment numbers. We exclude Serbian-majority municipalities because data reported are incomplete: Gracanice, Kllokot, Leposaviq, Mitrovice, Novoberdo, Ranillug, Pratesh, Sterpce, Zubin Potok, and Zvecan. 14 Overview Box 1. Improving the Effectiveness of the Social Assistance Scheme and its Potential Benefits to Education in Kosovo Globally, cash transfers targeted to poor families with children have a track record of protecting and improving children’s education, health and nutrition outcomes. Kosovo has a single social assistance program that provides regular cash transfer support to poor families: the Social Assistance Scheme (SAS).20 While SAS provides much needed income support to poor households, weaknesses in its design undermine its ability to support families to improve the education of their children. First, the categories of eligibility (Category I and II) restrict coverage of the SAS, disadvantaging poor families with children. Simulations based on the 2017 Household Budget Survey show that removing these eligibility criteria would significantly increase the number of poor children served by the scheme. Second, equally restrictive is the requirement for Category II SAS families to have at least one child under the age of five. While positive that families with young children are identified as eligible for SAS, there is no evidence to suggest that the cost of childrearing decreases after the age of 5. On the contrary, there is evidence of additional costs related to sending older children to pre-school and school. As a result of these restrictive eligibility requirements, the coverage of the SAS is low and SAS’s benefit incidence does not correspond to the country’s poverty profile. In fact, the SAS’s coverage rate is significantly lower than the country poverty rate and has been declining over time. Third, the SAS benefit amount increases quite slowly with family size and covers a lower share of the consumption needs of larger families compared with smaller ones.21 Finally, to be eligible for the SAS, households are required to show evidence that their children are enrolled in school. However, there is no evidence to suggest that schools or social workers follow-up to ensure that these children attend class throughout the school year. The Government has initiated a reform to the SAS to establish a poverty-targeted social assistance scheme. Delivering on this reform has the potential to better support children in poor and marginalized families to attend school and improve their learning outcomes. Source: World Bank. 2019. Kosovo Social Assistance Scheme Study. Washington, DC, 2019. Kosovo is struggling to provide quality education to all students from early childhood through secondary education. Environmental inputs such as caregiver or teacher stimulation, nutrition, or violence shape the architecture of the brain from the formative years onward and set the foundation for learning. Early learning indicators therefore provide good insight into the foundational skills required to master higher competencies. In Kosovo, only 23 percent of children aged 36–59 months were developmentally on track in the literacy-numeracy domain,22 which is much lower than in other countries in the region and in comparators (Figure 4). The percentage is even lower, just 7 percent, for children from the Roma, Ashkali, and Egyptian communities (UNICEF 2020). While foundational skills are low for all children, the learning divide widens for minority students as they grow. Only about 40 percent of all Kosovar children aged 7–14 years have foundational reading and numeracy skills. The proportions are even lower among children living in Roma, Ashkali and Egyptian communities with about 1 out of 5 children (18%) for reading and about 1 out of 10 children (13%) for math.23 Attendance rates for primary and lower secondary school are above 93 percent but drop to 90 for girls and 84 percent for boys in upper secondary. For children living in Roma, Ashkali and Egyptian communities, attendance rates are much lower, averaging 84 percent for primary, 63 percent for lower secondary, and 31 percent for upper secondary. Cultural 20 To be eligible for SAS, applicant families must fall into one of two narrowly defined categories: Category I – all family members are dependent and none is employed, and Category II - families have one family member who is able to work but must be registered as unemployed with the Employment Agency; all other family members must be dependent. In addition, the family must be parenting at least one child under the age of 5 years or providing permanent care for an orphan under the age of 15 years. Applicants also are subject to a poverty test. Beneficiary families of the SAS receive a child supplement for each child. 21 Beyond the second family member, the incremental benefit amount is only EUR 7.5 per month for adults and Euro 12.5 for children. As a result, the incremental benefit amount in response to an additional family member is smaller than that envisaged by the most common equivalence scales and does not reflect adult equivalent family size. 22 Children are identified as developmentally on track based on whether they can identify at least ten letters of the alphabet; read at least four simple, popular words; and know the names and recognize the symbols of all numbers from 1 to 10. Those who can correctly respond to at least two are considered developmentally on track. 23 Foundational reading skills are measured as the percentage of children age 7–14 who can: (1) read 90% of words in a story correctly, (2) answer three literal comprehension questions, and (3) answer two inferential comprehension questions. Foundational math skills are measured as percentage of children age 7–14 who can successfully perform: (1) a number reading task, (2) a number discrimination task, (3) an addition task and (4) a pattern recognition and completion task. Overview 15 factors such as early marriage may negatively affect school attendance at upper secondary. One in three women aged 20–24 living in Roma, Ashkali, and Egyptian communities reported being married or in a union before age 18 (UNICEF 2020). The quality of primary school education is low. Kosovo’s participation in TIMSS 2019 showed that a significant percentage of grade 4 students do not have basic proficiency in mathematics and science. In mathematics, 27 percent were unable to pass the minimum proficiency benchmark and in science the proportion increases to 41 percent. The percentage of students attaining minimum proficiency is considerably lower than comparator countries. The aspirational countries for Kosovo include several (Latvia, Lithuania and Czech Republic) that are nearing universal attainment of minimum proficiency standards (Figure 5). Figure 4. Children Aged 36–59 Months Developmentally on Track in Literacy-Numeracy, Percent Uruguay 81 2015 Albania 2005 N. Macedonia 2019 Moldova 2012 Kosovo 23 2020 Kyrgyz Republic 2018 Source: MICS UNICEF Figure 5. Fourth Grade Proficiency in Math and Science, Kosovo and Comparators, Percent Latvia Science Maths Czech Republic Lithuania Armenia Albania North Macedonia Montenegro Kosovo 59 73 Percentage of students attaining passing low benchmark Source: IEA 2019 (TIMSS 2019) By the end of lower secondary, about 79 percent of 15-year-old Kosovar students perform below the level of functional literacy in PISA 2018 (Figure 6). This is one of the highest percentages among all PISA-participating countries and much higher than the OECD average of 23 percent. Together with low overall quality of education, there are substantial achievement gaps across student subgroups. Students belonging to the poorest socio-economic quintile perform 45 PISA 16 Overview points behind student in the richest quintile (roughly equivalent to one year of schooling). There is some variation by region with Prishtina performing the best, at 28 points in reading (about three quarters of a year of schooling), better than the lowest-performing region, Gjakova. These results are worrisome for all economic quintiles. While students in Kosovo can expect to complete 13.2 years of schooling by age 18, when that number is adjusted for quality of learning, it is equivalent to just 7.7 years of school (HCI 2020). Figure 6. 15-year-old Students Performing Below Functional Literacy in Reading, PISA 2018, Percent OECD Average . % Estonia Slovenia Czech Latvia Lithuania Uruguay Moldova Montenegro Albania North Kosovo Republic Macedonia Source: OECD 2018. Similar regional variation within Kosovo appears in the results of the national Matura Exams. The Matura exams are taken every year by students in grade 9, at the end of lower-secondary and grade 12, the end of upper secondary. The Matura 12 exam is a high-stakes assessment that signifies both the end of a student’s formal secondary education and the gateway to higher or vocational education. In the 2019 Matura 12 exam, each student took three core subjects: native language, English and mathematics, and one subject chosen from 12 options. As the test design of Matura has evolved over time, 2019 results are not comparable to previous years limiting trend analysis. However, the results of the national assessment are comparable across regions. An analysis across municipalities reveal large differences in results. For example, while in Gjilan, 51 percent of the students passed in math and 65 percent passed in English, only 32 percent students passed in math in Hani I Elezit. Similarly, only 38 percent of students passed in English in Shterpce (Figure 7). Kosovo needs to have a system of learning assessments that guide teacher instruction and motivates improvement over time, especially in the lagging municipalities. Figure 7. Matura 12 Examination Results by Municipality, 2019 Percentage of students who passed Shtërpcë Malishevë Rahovec Shtime Kamenicë Novo Bërdë Lipjan Hani I Elezit Suharekë Gllogoc Kaçanik Dragash Viti Podujevë Skenderaj F.Kosove Klinë Ferizaj Junik Mean Obiliq Prizren Vushtrri Gjilan Gjakovë Mitrovicë Istog Prishtinë Deçan Native language English language Math Source: MESTI, Division for Standards, Monitoring, and Evaluation -Learning Assessments Unit Note: Small municipalities without graduates are not included. Serbian-majority municipalities do not sit for the Kosovar Matura exams. Overview 17 Students enrolled in vocational training perform worse than those in general education, and the gap is much wider in Kosovo than in OECD countries. For the PISA tests, students enrolled in academic programs outperformed students in vocational schools by 37 points in reading, 38 points in math, and 36 points in science (roughly equivalent to a year of schooling) [OECD 2018]. Even after accounting for the socio-economic profile of students and schools, students in general academic programs scored 35 points higher in science than students in vocational schools. Education continues to be of low quality even though Kosovo spends about as much on education as other middle-income countries. In 2019, Kosovo’s education spending equaled 4.6 percent of GDP and 16 percent of total government spending (Kosovo Boost Good Governance Data and KAS). This is similar to the expenditure by middle income countries that spent 4.3 percent of GDP and 16.3 percent of total government expenditures in 2018 (average of 35 countries in upper-middle and lower-middle income category with data available for 2018). In comparison to Kosovo’s structural and aspirational peers, Kosovo spends a comparable percentage of GDP and total government expenditure on education (Figure 8). Over the last decade, Kosovo’s education spending has almost tripled (Figure 9). Of the total government expenditure on education, 71 percent is spent on pre- university education: 55 percent on pre-primary, primary, and lower secondary, and 16 percent on upper secondary). Figure 8. Kosovo’s Spending on Education Compared, Percent of Total Government Spending and Percent of GDP Percentage (%) . . . . . . . . . . . Percentage of . total government . . . . . . expenditure . . . Percentage of Albania Armenia Lithuania Kosovo Latvia Uruguay Estonia Kyrgyz GDP Republic Source: UNESCO Institute of Statistics Figure 9. Education Spending, Kosovo, 2007–19, Euros Expenditure (Euros) Millions Education, n.e.c. R&D education Subsidiary services for education Education, level undefined Tertiary education Post secondary non-tertiary education Upper secondary education Pre-primary, primary and lower secondary education Education Source: World Bank BOOST data for Kosovo, 2007–19. Note: Of the education budget line that is undefined, 30 percent went to the Kosovo Academy for Public Safety, 15 percent for educating persons with special needs, and 12 percent on professional training and training of teachers. Spending classified as Education, n.e.c, supports local education departments. 18 Overview In Kosovo’s decentralized system, municipalities play a large role in delivery of preprimary, primary, and secondary education but resources are quite limited as most spending goes to covering the wage bill. Municipalities are responsible for registration and licensing of educational institutions and recruitment, payment of salaries, and training of instructors and administrators (Law No.03/L-068), while major construction projects for educational institutions can be funded by direct grants from the MESTI. The MESTI is also responsible for overall strategic guidance, standard-setting, expansion of access to education services and quality assurance. As a result, 84 percent of spending on pre-university education is executed by local governments and only 16 percent by the central government. Although spending on pre-university education has more than doubled over the last decade, the major growth has been primarily for wages and salaries, while spending on goods and services has gone up only slightly (Figure 10). In 2019, wages and salaries constituted 89 percent of total local spending on pre-university education, while goods and services constituted 7 percent, capital expenditures 3 percent, and municipal expenditures/utilities only 1 percent. In 2019 a much larger percentage of central expenditures on pre-university education went to capital expenditures (42 percent) and goods and services (39 percent). Figure 10. Spending on Pre-university Education by Category, 2009–19, Euros Expenditure (Euros) Millions Local Capital expenditures Subsidies and transfers Municipal expenditures/ Utilities Goods and services Wages and salaries Central Capital expenditures Subsidies and transfers Municipal expenditures/ Utilities Goods and services Wages and salaries Source: World Bank BOOST data for Kosovo 2009–2019. General government funds account for 98 percent of local spending on pre-university education, with the remaining 2 percent drawn from municipal revenues. The Law on Local Government Finance, no. 03/L-049, 13 March 2008, and the Law on Public Financial Management and Accountability, no. 03/L-048, 13 March 2008, made the central government responsible for providing local governments with specific education grants through the Grants Commission to finance the cost of providing a minimum standard of compulsory education. Provision is based on such parameters as student enrollment and pupil-teacher ratio for allocation of personnel and payroll by calculating the average remuneration, funding for replacement of teachers on medical or maternity leave, funding of educational and training institutions by type and size of facilities, funding for schools in deep mountain zones, and funding per pupil for operating expenses and capital investments. Local governments can supplement these specific grants with revenues generated by educational and training institutions and/or with additional funding from the general municipal grant, depending on local financial opportunities. However, only in Prishtina do municipal revenues make up as much as 10 percent of total pre-university spending. Domestic and international donations and grants constitute only 0.1 percent of local government spending on pre- university education. Spending on wages and salaries in the education sector has almost doubled over the past decade. Since the number of teachers has not substantially increased, the heightened spending is primarily driven by increases in public wages. Moreover, there has also been no decrease in the number of teachers in tandem with the declining student population or migration from rural to urban areas. Though salaries of teachers are slightly lower than for other professional occupations (Figure 11), no indication of dissatisfaction with the amount of salary is reported in teacher surveys. Overview 19 Figure 11. Kosovar Salaries in the Education Sector and Other Comparable Sectors, 2018 Information and Communication Mining and Quarrying Public Administration and Defence; Compulsory Social Security Human Health and Social Work Activities Financial and Insurance Activities Professional, Scientific and Technical Activities Water Supply; Sewerage, Waste Management and Remediation Activities Education Construction Transportation and Storage Administrative and Support Service Activities Real Estate Activities Activities of Households as Employers; Undi erentiate Goods and… Wholesale and Retail Trade; Repair of Motor Vehicles and Motorcycles Arts, Entertainment and Recreation Manufacturing Agriculture, Forestry and Fishing Other Service Activities Accommodation and Food Service Activities Monthly Salary (Euros) Source: KAS 2018. Though grants from the central government follow an agreed formula, there are wide differences between overall budget per enrolled student across the municipalities (Figure 12). National average expenditure per enrolled pre-university student is ¤757 annualy. However, there are huge differences across municipalities, ranging from ¤701 in Fushë Kosovë to ¤2,245 in Kamenicë. The per-student costs reflect student-teacher ratios across municipalities (Figure 13). Municipalities like Kamenicë (1:10), Dragash (1:12) and Gjilan (1:13) have the lowest student-teacher ratios and therefore the highest per-student costs. (By law the student-teacher ratio is set at 1:21 for non-mountainous areas.) There is a strong need to add equity measures to per-capita funding across municipalities and increasing spending efficiency to improve education outcomes. 20 Overview Kamenicë . Fushë Kosovë Dragash . Hani i Elezit Figure 13. Figure 12. Gjilan . Prishtinë Viti . Prizren , Source: KAS and MESTI. Deçan . Gllogoc , Malishevë . Mamushë , Obiliq . Pejë , Junik . Mitrovicë Jugore , Source: World Bank BOOST Data 2019 and KAS. Rahovec . Obiliq , Student-Teacher Ratios by Municipality Kaçanik . Junik , Suharekë . Podujevë , Shtime . Malishevë , Pejë . Rahovec , Prizren . Kaçanik , Gjakovë . Vushtrri , Per-student Costs for Pre-university Education by Municipality Istog . Klinë , Mitrovicë . Ferizaj , Total . Shtime , Gllogovc . Lipjan , Klinë . Suharekë , Mamushë . . Istog , Skënderaj . Gjakovë , Ferizaj Lipjan . Skënderaj , Podujevë . Deçan , Prishtinë . Viti , Vushtrri . Gjilan , Hani i Elezit . Dragash , Fushë Kosovë . Kamenicë , Overview 21 02 Improving Education Quality Large variations in primary and secondary class sizes and an average student-teacher ratio much lower than the 1:21 guideline leaves little room after the wage bill for investments to improve the quality of education. In 2019 while there are about 35/36 students per upper secondary class in Fushë Kosova; there was only 1 student per class in Leposaviq. In general student-teacher ratios are low in both primary and secondary education (Figure 14) and is much lower than the stipulated in AI MESTI 22/2013.24 Like class sizes, student-teacher ratios are also expected to vary widely across municipalities (Figure 13). The variation in class-size ratios highlights the need to optimize schools and teachers, as discussed below in sections IV and V. Figure 14. Student-Teacher Ratios in Kosovo, 2014–19 Number of teachers Upper secondary , , , , , , Pre-primary, primary and lower secondary Pre-school , , , , , , Students per teacher Pre-school / kindergarten - STR Primary / lower secondary - STR Upper Secondary - STR Pre-primary STR Source: MESTI Teachers in Kosovo enjoy a collegial work environment that can be a strong foundation for Kosovo’s teacher professional development program. Teachers are predominantly satisfied with their jobs. According to research conducted by MESTI in 2016 involving 303 teachers, about 94 percent agreed that “the atmosphere in the teachers’ room is good,” and even more agreed that the environment with other teachers and school director was supportive and cooperative. Similarly, 84 percent agreed with the statement “I am appreciated by my colleagues.” These results indicate a healthy work environment conducive to a strong school-based professional development and training program. In fact, certain features that can be a foundation of a strong teacher professional development program are already in place. For example, survey results show that school management regularly conducts in-class observations (82 percent) and that teachers have the opportunity to discuss needed improvements. Similarly, of the teachers who attended professional development activities, about half attended school-based reflection and planning sessions monthly. However, gaps persist. For example, satisfaction is low among certain sub-groups of teachers, especially young teachers, and around half of the teachers participating in professional development attended workshops and activities outside the schools that might not have been fully aligned with the school instructional plans. Kosovo is prioritizing teacher development and is introducing reforms to improve teacher quality, however, resources at the school level are not available to implement new strategies learned. A five-stage teacher licensing scheme is in place nationally and teachers are required to participate in continual professional development and undergo performance assessments that determine their job continuation and promotion. However, this legal framework has not yet been implemented and professional development activities are still conducted ad hoc through 24 Article III of the Administrative Instruction states that the student-teacher ratio for all levels of primary, lower, and upper secondary is 1:21.3 for the majority community and 1:14.2 for minority and mountainous communities. If a special education student is present in the class, then from the maximal number, two students should be removed. 24 Improving Education Quality workshops and seminars. Teachers felt that the resources and circumstances in their schools were not supportive of the strategies learned in professional development workshops (MESTI 2016). Teachers also emphasized the need to develop skills in special needs education, understanding new curricula, and inter-curricular skills. It is thus imperative to develop school based professional development mechanisms and processes to complement the central driven program. Classroom observations are required to further understand the realities in schools and design a professional development program that supports teachers in delivering quality education. KESP 2017–21 also emphasizes the need to develop coordinated standards for “pre-service” and “in-service” teacher education. There is a strong need to prioritize investment in quality-enhancing school inputs to support teachers and schools in delivering quality education. While schools do not lack teachers or support staff, 86 percent of 15-year-old students (as assessed by OECD 2018) study in schools where school’s instruction was hindered by a lack of educational materials like textbooks, IT equipment, library or laboratory material (Figure 15). Administrative data reveals similar problems. For example, according to this data, only about 50 percent of Kosovo’s schools have a library which confirms the need for students to have access to books to engage in informational and recreational reading to strengthen their functional literacy skills. In recent years, a great effort has been made for schools to access broadband and only about 120 out of about 1,200 have yet to receive access. While connections have been made, availability of IT equipment is very low and varies widely across the municipalities. Fushë Kosova, for example, averages only one computer for 85 students. Similarly, about 50 percent of students are in schools where lack of and poor quality physical infrastructure hindered educational instruction (Figure 15). Addressing infrastructure deficiencies in primary and secondary education requires sizable investments. Of the investment needs estimated in 2016 at ¤318.6 million (5.2 percent of GDP), 55 percent of the total, ¤176.6 million, was needed for new construction, which includes new school spaces, such as annexes, and 45 percent, ¤142 million, was for rehabilitation of existing infrastructure.25 Table 1 provides a snapshot of the state of school facilities as assessed in 2016. More than 83 percent of basic education facilities lacked potable water or had dysfunctional sanitation and 50 percent of structures were considered unsafe. Table 1.  School Facilities in Kosovo With Inadequate Infrastructure by Type of Problem, 2016 Number of school facilities with inadequate infrastructure School Infrastructure Inadequacies, Criteria 1-9 Basic Education Upper Secondary C1- Unsafe & depreciated structures, nonfunctioning electricity 528 45 C2-Non-drinkable water or dysfunctional sanitation 882 111 C3-Facilities >50% depreciated or “shed schools” 17 0 C4- Three teaching shifts and overcrowded classes (>36) 37 24 C5-Facilities without laboratories 757 107 C6 -Facilities without labs and IT equipment 915 101 C7-Facilities without access for the disabled; no emergency exits 813 87 C8-Facilities with low energy efficiency 961 102 C9-Facilities lacking sports or recreational infrastructure. 977 119 Total number of facilities 1,053 164 Source: Data from School Map Report 2016, World Bank staff calculations. Note: The total number of school facilities was 1,217 in 2016; basic education = 1,053 and upper secondary = 164. 25 Of this amount, 36%, or €114.3 million were estimated as high-priority investments, which deal with infrastructure deficiencies regarding that threatened the structural safety of educational facilities, provision of healthy areas, replacing of container schools and over-depreciated facilities, elimination of three shifts and classes with over 36 students, and increasing the number of laboratory classrooms. The other 64% or €204.3 million was considered medium priority, such as provision of research and laboratory equipment, facilitating access for persons with disabilities, increasing energy efficiency of educational facilities, and expanding the sports and recreational infrastructure. Improving Education Quality 25 Figure 15. Percent of Students whose School Principals Responded “To some extent” or “A lot” to the Question: Is your school’s capacity to provide instruction hindered by any of the following issues? Inadequate or poorly qualified teaching sta 10.2 . Kosovo OECD Inadequate or poorly qualified assisting sta 16.6 . A lack of teaching sta . 19.1 . A lack of assisting sta . 28.4 . Inadequate or poor quality physical infrastruce 48.5 . A lack of physical infrastructure 50.4 Inadequate or poor quality educational materials 80.4 . A lack of educational materials 85.9 . Percentage of students Source: OECD 2018. 26 Improving Education Quality Improving Education Quality 27 03 School Optimization Scenario for Pre- University Education The demographic decline has resulted in oversized school networks and low student-teacher ratios. However, urban migration has also led to overcrowded schools and continued double shifts, particularly in upper secondary, where school populations average 1,800 students. The lack of targeted approaches to channel resources to underfunded levels of education like ECE or to schools that are falling behind exacerbates the inequities. Optimizing the school network may allow freeing up of public financial resources to improve education quality and efficiency. Reducing the number of school facilities in the network lowers fixed costs linked to running these facilities, like electricity, heating, rent, and housekeeping. Fewer facilities also allow for consolidating small classes and therefore thus lower the number of staff positions needed, mainly teachers but also management and support positions. After an adjustment period, as transportation may need to be set up and staff positions right-sized through retirement, retraining, or transfer, savings can be reallocated to invest in educational materials and improved learning environments. School optimization in Kosovo might be difficult to implement in practice, hence, it should be carefully planned. Based on the data available, there seems to be room for a number of school mergers. However, many of the schools which may currently be too small and inadequately equipped for efficient learning may be located in socially deprived, remote, or mountainous areas. Closing down such facilities could have an adverse impact on the social life of these remote villages. It may be necessary to provide public transportation for students so that they do not have to walk for an unreasonable amount of time to attend school. The costs and technical arrangements may be challenging for municipalities already struggling to provide basic amenities such as toilets in all schools. Adequate resources should be budgeted and provided to ensure continuity of education. Larger schools also favor learning efficiency. School consolidation allows for larger classes where students have access to better learning, particularly for grades 6–12 when specialized subject teachers are needed. It tends to reduce clustering of students from different grades in the same class, which can reduce learning effectiveness if teachers are not well trained for multi-grade teaching. It also improves the quality of facilities by making it possible to close the less suitable ones. It can also help reduce the spread of teaching positions over multiple units, for example for subject specialists who are assigned more than one location and improve the physical access to schools. In Kosovo the links between school institutions and school facilities are not straightforward. There are about 753 school institutions in Kosovo, which are legal entities, assigned a name, a management, and a staff. A given school institution can, however, have several school units. Each school unit is either the main school or a subsidiary named satellite school. In 2016, according to the School Map Report,26 there were about 1,200 such school facilities operating in Kosovo. Some of these are clustered so the actual number of school sites is smaller. Since 2016, some of these facilities have been closed down, so the current number of facilities may now be lower.27 In 2019, there were about 1,072 school units in Kosovo, of which about 300 were satellite schools, according to the latest data from the National Education Management Information Systems (EMIS/SMIA) data. There is not always a perfect match between a school site and a school unit because some school facilities (buildings) may be shared between schools while others may have been closed. Such a discrepancy complicates the computations of potential school optimization as some school units cannot be localized on the map, and it may be impossible to compute the number of available classrooms in some facilities without onsite verification. Overall, there are three types of school units welcoming children aged 9 months to 18. Out of 1,072 school units, they are broken down as follows: (1) about 728 (70%) are integrated schools consisting of students from grade 0 to 9 (primary and lower secondary); (2) about 169 (16%) enroll only primary students (grades 0 to 5), and (3) about 118 (11%) enroll only upper secondary students. In addition, about 40 school units enroll young children (nursery/preschools, 4%), and less than 20 units either enroll all types of students from grade 0 to 12 or only lower secondary students (Figure 16). Satellite schools are fairly small and have fewer than the minimum threshold of students per school. Most of the primary only school units (93%) and 15 percent of integrated primary and lower secondary school units are satellite schools. The size of the satellite primary schools is quite small, with only 43 students on average (less than 8 per grade); the main primary schools average 290 students. This is well below the legal minimum threshold of 84 students per school, for which construction of a new facility may be authorized. The same holds true for integrated primary-lower 26 School Map Report, National Level, 2016. Inventory of Educational Facilities in Kosovo. 27 As of August 2020, an updated database of school facilities had not been made available to the Bank. 30 School Optimization Scenario for Pre-University Education secondary schools, which have about 100 students in satellite units (10 students per grade), whereas main units have 370 students. This is also well below the legal threshold of 210 students triggering the building of a new facility of that type. For upper secondary schools, almost all main school units have more than 100 students, allowing at least one full class of 30 students per grade. Most units below that threshold are satellite units at this level of education as well. Figure 16. Distribution of the Types of School Units in Kosovo, 2016 Upper All levels secondary only Creches Lower secondary only Primary only Primary and lower secondary Source: School Mapping Report 2016 More than half of primary and lower secondary school units are smaller than the local norm would consider efficient. Using the legal thresholds of 84 students for primary school units and 210 for integrated primary-lower secondary school units, about 89 percent of primary school units (about 150) and 51 percent of integrated primary-lower secondary school units (376) could be deemed “small.” These statistical observations suggest that Kosovo’s school network could benefit from optimization to channel savings into investment in areas that improve education quality. Beyond the mere size of the school unit, several other factors affect whether school network optimization is appropriate: • The physical condition and current features of existing facilities, which affects whether those with room to welcome more students are actually suitable; • Local demographic trends, which suggest whether a local drop in student population is temporary or lasting; • The availability of transportation to relocate students from smaller schools to receiving ones; • The social value of a school facility to maintain activity in villages and small towns—families may decide to leave the locality if the local school unit is closed; • Local preferences for smaller class size and shorter distances versus the costs saved for the community budget; • The availability of spare classrooms allowing unit mergers within a reasonable radius; and • Current regulations about teacher-pupil ratios and the maximum number of students per class. School Optimization Scenario for Pre-University Education 31 Designing an optimal network that can be implemented effectively requires a very detailed set of information, not only about school infrastructure, but also about enrollment, staffing, actual costs, and local legal and political constraints. One principle guiding school network optimization is whether students of small school units can be accommodated within the empty classrooms of a larger facility nearby. This requires using school and facility-level data to compute: (1) the number of empty classrooms in each facility that could be used to host students from a nearby consolidated school and (2) the number of classes in each unit to be fitted into a nearby school. There can be many constraints for both consolidated and receiving facilities. Among them are likely to be (1) a maximum distance between the two merged school units, which could in turn depend on local conditions, such as the altitude; (2) adequate conditions at the receiving facilities; (3) adequate transportation for relocated pupils; (4) maximum number of students per class; and (5) student-teacher ratios. The effects of distance and regulatory constraints on the possibilities of optimization can be explored using a dedicated algorithm. The algorithm’s main principle is to fit the smallest school units into the facilities with the most empty space (Figure 17). (For detail on the algorithm and data constraints see Annex 2.) Figure 17. Principle of the School Network Optimization Algorithm School units are School facilities are ordered by size ordered by # of empty classrooms Random Draw Receiving school facilities of similar vacant capacity are also jumbled Small school units are integrated into empty school facilities until there is no school unit large enough to receive it Source: Author’s illustration Pairing data about facilities and enrollment by grade at school units makes it possible to estimate the number of class groups and empty classrooms. The theoretical number of groups can be easily estimated from the maximum number of students allowed per class, assuming that no more than two distinct grades can be taught in the same group. The number of empty classrooms is the number of classrooms in the facilities minus the number of class groups computed as indicated above. 32 School Optimization Scenario for Pre-University Education School Optimization Scenario for Pre-University Education 33 04 Simulation of network optimization in the municipality of Suharekë To explore the possibilities for school consolidation and savings, school network optimization was simulated for one municipality in Kosovo. Suharëkë was chosen due to its size, location, and characteristics similar to other municipalities. Like most of Kosovo beyond the capital, Sukareke is rather rural. It has 46 schools with 51 distinct school units/facilities of 39 at regular altitude (300– 750m) and 12 in areas above 751m. Like most municipalities, Suhareke delivers education in shifts, with 20 schools offering 2 shifts and one school 3 shifts. Suharëkë also has three preschools and three upper secondary schools. The upper secondary schools each offer a different specialization so that there is no obvious mapping in this municipality between lower secondary and upper secondary schools. There are 5 very small primary schools, 28 schools integrating grades 0 to 9, and 7 serving grades 1 to 9. Half of the integrated primary-lower secondary schools have fewer than 210 students, which is the legal threshold for opening a new facility and could therefore be considered “small” (see Figure 18). Figure 18. Map of School Units in Suharekë IP Malisheve P.F. P.F. + Sh.F. Shtime P.F. + Sh.F. + Sh.M.U. Sh.F. Sh.F. + Sh.M.U. Sh.M.L. Gjimnaz Sh.M.L.Mx Sh.M.L.P. Teknike IAL Ferizaj Mamushe Prizren Shterpce Source: Appendices of the School Mapping Report 2016, Municipality of Suhareke, p.3, Annex 1: Municipal map of the spatial distribution of educational institutions. Note: Upper secondary schools in purple, beige and deep blue, are clustered in the central urban area. There is room in Suharëkë for some school consolidation. There are 40 school units in Suharëkë, excluding the preschools and the upper secondary schools. In 27 there may be a total of 60 to 80 empty classrooms, although some of these may not be in adequate condition. School optimization possibilities increase and converge rapidly with the distance permitted between merged schools.28 The state of infrastructure does not limit optimization because larger schools are relatively better equipped than smaller ones, but it underlines that savings through optimization could clearly help improve and update the existing infrastructure. The cost of improving the physical facilities at all 45 schools without any consolidation of facilities was estimated at ¤12,096,498, or 0.2 percent of GDP in 2016.29 28 One can run the algorithm 1,000 times for each maximum distance and retain the best solution within the acceptable distance. 29 Construction Management and Consulting (CMC). 2016. Capital Investment Projections in Educational Facilities, National and Municipal Level. 36 Simulation of network optimization in the municipality of Suharekë Table 2.  School Facilities in Suharekë, by Type of Infrastructure Problem, 2016 Number of school facilities with inadequate infrastructure School Infrastructure Inadequacies, Criteria 1-9 Basic Education Upper Secondary C1- Unsafe & depreciated structures, non-functioning electric 21 1 C2-Non-drinkable water or dysfunctional sanitation 36 3 C3-Facilities >50% depreciated 0 0 C4- Three teaching shifts and overcrowded classes (>36) 1 1 C5-Facilities without laboratories 33 2 C6 -Facilities without labs & IT equipment 37 2 C7-Facilities without access for the disabled, no emergency exits 33 2 C8-Facilities with low energy efficiency 41 3 C9-Facilities lacking sports/recreational infrastructure. 41 3 Total number of facilities 42 3 Source: School Map Report 2016 data. To estimate possible savings from optimization for Suharëkë, the simulation used three typical scenarios (see Table 3). In the limited scenario, only schools below 700m in altitude can be consolidated, with a maximum distance of 3 km between schools. Student-teacher ratios of 19.3 and 14.3 (for schools above 700m of altitude) can be applied to simulate the number of teaching positions. In that configuration, about two facilities could be consolidated and about 1 of 24 teaching positions could be rationalized (see Table 3). In the medium scenario, only schools below 700m can be consolidated, with a maximum distance of 4 km between schools. The regulatory teacher-pupil ratios are applied to estimate the number of positions. About 3 facilities could be consolidated and 1 of 9 teaching positions could be rationalized. In the ambitious scenario, all schools within 5 km could be consolidated. The regulatory teacher-pupil ratios are applied to estimate the number of positions. Up to 12 facilities could be consolidated and 1 of 6 teaching staff could be rationalized. Such a scenario would require providing transportation to the students, which would reduce the net financial savings, as students cannot be required to walk more than a few kilometers, especially in mountainous areas. However, school consolidation might also provide savings through the capital investments foreseen to improve the existing infrastructure. Table 3.  Estimated Savings for Suharekë in Three School Network Optimization Scenarios 1. Limited Scenario 2. Medium Scenario 3. Ambitious Scenario Only schools below 700 m altitude can be Schools < 700 m altitude with maximum distance All schools within a maximum distance of consolidated, with a maximum distance of 3 km of 4km can be considered for school mergers. 5km can be considered for school mergers between schools. Apply pupil/teacher ratios of 19.3:1 and 14.3:1 Apply regulatory pupil/teacher ratios (21.3:1 and Apply regulatory pupil/teacher ratios (21.3:1 and according to altitude 14.3:1 according to altitude) 14.3:1 according to altitude) Savings: 1 of 20 facilities, 1 of 24 staff Savings: 1 of 13 facilities, 1 of 9 staff Savings: 1 of 3 facilities, 1 of 6 staff Estimating financial savings from optimization would require detailed data at the school level. Being able to estimate financial savings would require access to detailed data on the staff, their compensation, and how much each specific school spends on operations. Although there may be fixed costs, operating costs of facilities are likely to differ according to their size, age, level of occupancy, or the altitude (schools in higher altitude may require more heating in winter). In the short run, the staff savings might be limited as teachers may not be laid-off and municipalities may prefer to rely on retirements or provide incentives for early retirement. As a consequence, municipalities would need to manage expectations and face potential political and social costs without initial rapid financial benefits. School optimization may also require transition costs, related to renovation or extension of facilities used as hubs after the merger. In the medium to long term, however, municipalities would experience savings from school network optimization that could make it possible for municipalities to invest a larger proportion of school grants in areas that improve school quality such as teaching and learning materials. Simulation of network optimization in the municipality of Suharekë 37 05 Skills Gap Discussion The low level of skills development in the early years continues on to higher levels of education and as a result, adults in the country have low literacy skills. Literacy is critical because it serves as the basis for further skill acquisition. The core literacy test in the Household Survey conducted as part of the STEP Employer and Household Surveys provides an indication of adult basic literacy. In Kosovo, nearly one in seven respondents (14 percent) failed the core literacy test. For those who continued to the full literacy test, the average score remained low, barely reaching level 2 of 5, indicating that adults in Kosovo can recognize basic vocabulary, evaluate the meaning of sentences, and read paragraph text; however, they have limited ability to understand and evaluate dense and complicated texts that combine relevant and irrelevant information or that require background knowledge. These results illustrate the poor quality of overall education in the country compared to other Western Balkan countries in the region (Figure 19). Figure 19. Mean Literacy Scores, Total Working Age Population and Employed (aged 15-64), Western Balkans, FSU, and OECD Countries Total population Employed Kosovo Georgia Armenia Serbia Ukraine Czech Estonia Slovak OECD Republic Republic average STEP PIAAC (OECD) Source: STEP Household Surveys (Kosovo 2018, Serbia, Armenia, Georgia, Ukraine), Programme for the International Assessment of Adult Competencies, PIAAC (Others) Together with low average literacy scores, there are significant differences in foundational skills across different groups. Results from the core literacy test suggest that a few groups fall significantly behind the national average in terms of basic literacy. This is especially true for individuals belonging to ethnic minorities other than Albanian or Serbian (Turks and Roma are the biggest minorities after Serbs). Among ethnic minorities other than Albanians or Serbians, half of the respondents failed the basic core literacy test and so did not continue to the full literacy test (Figure 20). Though literacy in Kosovo does improve with more education, 40 percent of adults with tertiary education were unable to pass the full literacy test (Level 8) (Figure 21). Figure 20. Distribution of Core Literacy Scores by Demographics, Percent, 2018 % % % % % % Total Male Female - - - - - Albanian Serbian Other Gender Age group Ethnicity Fail ( - ) Pass ( - ) Pass ( - ) Pass ( ) Source: Kosovo STEP Household Survey, core literacy assessment, 2018 40 Skills Gap Discussion Figure 21. Distribution of Core Literacy Score by Education and Labor Market Status, Percent, 2018 % % % % % % Lower secondary Upper Tertiary Employed Unemployed Students Other inactive or less secondary Education Labor market status Fail ( - ) Pass ( - ) Pass ( - ) Pass ( ) Source: Kosovo STEP Household Survey, core literacy assessment 2018. Employers’ views confirm the presence of quality problems in the education system. According to the World Bank Survey of households and employers—Skills Toward Employability and Productivity (STEP),30 25 percent of firms consider skills shortages a major obstacle to growth; 75 percent of firms that attempted to fill a higher skill position and 60 percent of firms that tried to fill a medium to lower skill position encountered problems because of shortages of either skills or experience. The skill gaps were consistent across all major occupational categories. Among the medium to lower skills positions, fewer firms experienced difficulties recruiting for elementary occupations or jobs requiring driving or operating machinery (lower skill occupations) than for clerks, sales, services workers, or construction workers. Education in Kosovo gives students limited practical experience and professional skills. Nearly 60 percent of employers interviewed in the STEP survey thought that the general education system did not provide relevant practical experience to students and nearly 50 percent thought that the education system did not provide up-to-date knowledge, or socio-emotional skills like discipline, timeliness, or people skills. Although employers are slightly happier with TVET outcomes, 35 to 42 percent thought that the training systems do not provide people with necessary skills. This is a serious concern, since the objective of TVET is to provide students directly with work relevant skills in a profession or vocation. Kosovo is also struggling with serious labor market challenges. Less than 33 percent of adult Kosovars have a job, almost nine out of ten women are not working, and over half of active youth are unemployed. There are relatively few jobs in the market and informal employment is high. In the last three years, for each occupational category, only 15 to 30 percent of all firms in the Employer Survey (conducted by the World Bank in 2018) tried to hire at least one person. The final sample was 500 workplaces (484 firms and 16 branches) Globally, technological and organizational changes are transforming jobs and the corresponding skills in demand. Significant structural change, in particular the rising role of information and communications technology in the production of goods and services, has led to a decline in the need for routine manual skills and an increased demand for “new economy skills” that complement automated work (Autor, Levy, and Murnane 2003). Structured and repetitive tasks are giving way to tasks that require abstract thought, decision making, teamwork, and leadership. Manual tasks now require adapting and reacting to changing circumstances with tools and manual dexterity, as well as social interaction and collaboration. Thus, workers must accumulate an array of general skills that form the basis for acquiring job-specific skills and adapting to changes in tasks over time. In Kosovo, according to the STEP Survey, demand was greater for higher-skill occupations (managers, professionals, technicians) and medium-skill service-related occupations (clerks, sales workers and 30 The STEP survey consists of two surveys: (1) The Kosovo STEP household survey with a target population of noninstitutionalized persons aged 15–64 living in private dwellings in urban areas of the country, excluding foreign diplomats and non-nationals. The final sample was 3,511 individuals and the survey took place from September 2016 through March 2017. (2) The Kosovo STEP employer survey targeting establishments, or workplaces, defined as distinct physical locations at which employers undertake economic activity. The sample was stratified by two geographic domains (Prishtina and the rest of Kosovo) and four sizes of employers by number of employees: 5–9, 10–15, 16–50, and 51+ employees. This survey took place from November through December 2015. Skills Gap Discussion 41 services workers) than for low skill and more traditionally male-oriented occupations (construction and crafts, elementary occupations, drivers, and operators).31 A broad set of skills, both cognitive and socio-emotional, is needed to fulfill jobs in Kosovo. The STEP Household Survey found that cognitive skills, both foundational (numeracy, literacy) and higher-order (learning and problem solving), are used on the job, to a medium or high degree by between 40 to 60 percent of workers, depending on occupation. Among socio-emotional skills, interpersonal skills when working with clients and team members, are an important part of the job for nearly 70 percent of workers. Similarly, according to the STEP Employer Survey, employers confirm that jobs overall are skill-intensive, using a broad range of cognitive and socio-emotional skills. A majority of workers reported that their education levels were well matched with their jobs, but not those with post-secondary non-tertiary education. In Kosovo, more than 65 percent of workers report a good match between their education and job requirements, 23 percent consider themselves overeducated, and 11 percent undereducated (Figure 24). This is broadly aligned with the pattern seen in comparator countries like Serbia or North Macedonia (Figure 25). Across all levels of education, most workers report being well matched, except for post-secondary, non-tertiary education, where a majority consider themselves undereducated for the job. This is important because a large percentage of students in Kosovo enter vocational education. Figure 22. Worker’s Perceptions of How Their Education and Skills Match Their Jobs Education and jobs matching All Tertiary Post-sec, non-tert Upper sec Lower sec % % % % % % undereducated wellmatched overeducated Source: STEP Household survey 2018. 31 Women were overrepresented among professional occupations, and underrepresented among the construction and crafts, and plant and machine operators and assemblers. 42 Skills Gap Discussion Figure 23. Education and Jobs Matching, Kosovo and Comparators Education and jobs matching, Kosovo and comparators Kosovo Serbia Ukraine FYR MK Georgia Armenia % % % % % % undereducated wellmatched overeducated Source: STEP Surveys. Firms are not strategically connected with education and training systems. Connecting employers, workers, education systems and students is essential to make school-based training more relevant. Collaboration can range from public-private partnerships with shared financing and management responsibilities, to providing technology/equipment, jobs and skills needs assessments, mentoring and career advice, partnerships for curricula reform, and work-based learning (Dunbar 2013). However, fewer than 20 percent of firms in Kosovo have regular contacts with the education system for internships or employee trainings. For example, 30 percent of vocational schools in Kosovo specialize in trades for which there is little demand in the market (KESP 2017–21). As a result, a large number of youth in Kosovo have educational qualifications that are not sought after by employers. The broad set of cognitive and socio-emotional skills that are valued in the labor market must be nurtured from an early age, throughout the education system, and beyond. Labor-market- relevant skills include the transversal cognitive and socio-emotional skills that are necessary for modern, competitive firms, such as conscientiousness, independent and creative work, ability to work under rapidly changing or stressful conditions, collaboration with others, and communication. These are developed from birth and throughout life: from ECD programs that lay the foundation for basic cognitive and socio-emotional skills, through school-based academic or vocational learning and into university, then through training and retraining programs for adults also skills development in the work place. Education and training systems cannot be expected to provide fully formed workers. Their responsibility is to produce trainable graduates that have the necessary foundations to be further trained on the job and adapt to new challenges as they arise. Stepping up skills development in Kosovo will therefore require collaboration between all stakeholders: education systems, firms, students, and policy makers. Skills Gap Discussion 43 06 Impact of COVID-19 on the Education System As in most other countries around the world, the education system in Kosovo has been seriously affected by COVID-19. The pandemic forced Kosovo to close schools from the second week of March 2020 through the end of the school year, 30 June 2020, and transition to emergency remote teaching using online platforms and TV-based broadcasts. However, due to the emergency nature of remote teaching initiatives, learning effectiveness is likely to be lower than that of traditional classroom-based instruction. Moreover, many teachers and students were unable to participate in distance learning or communicate effectively due to lack of an Internet or wifi connection and/or access to a digital device (smartphone, tablet, laptop, computer), because they had none, or because they had to be shared within a family. The start of the 2020–21 school year was delayed and had a phased face-to-face opening. COVID-19 is still affecting education delivery; at any time a classroom or school may be closed or quarantined because of exposure to the virus. In Kosovo, the COVID-19 pandemic is expected to result in substantial learning losses and push even more students into functional illiteracy. Assuming that a student gains 40 PISA points of learning in a year, that schools are closed for an average of four months, and remote teaching in Kosovo is 45 percent as effective as face-to-face teaching, we estimate that learning in Kosovo will drop by the equivalent of 9 PISA points in reading (Figure 24). Additionally, while most students in the country already perform below the threshold for functional literacy and assuming some will lose more than others, such learning losses imply that the percentage of students performing below functional literacy may further increase by 1 percentage point (from 79 to 80 percent). The differential access to and effectiveness of remote teaching is widening learning gaps across student populations. While remote teaching is likely to be less effective for all students, it is even less effective for students from lower socio-economic quintiles who might suffer from poor connectivity and limited access to electronic devices. Also, regardless of income, families with several children have the added challenge of balancing parental and child needs for connectivity and devices. Students with special needs may also have to deal with unique challenges accessing remote learning content. Using access indicators from PISA 2018 data and assumptions on differentiated learning effectiveness (25 percent effective for the poorest, 50 percent effective for the average, and 75 percent effective for the richest students), the reading achievement gap is expected to increase by 18 percent (from 45 to 53 PISA points) after only a short-term school closure (Figure 25). Figure 24. Estimated Impact of COVID-19 on PISA Scores PISA Scores in Reading KOS XKX estimate Post-COVID Note: About learning loss estimates: The World Bank has estimated the effects of COVID-19 related school closures on learning outcomes for 157 countries. Simulations use data on learning outcomes, years of schooling and monthly wages to estimate the potential effects of school closures in general and across socioeconomic groups. For global and regional estimates, see World Bank 2020a. 46 Impact of COVID-19 on the Education System Figure 25. Estimated Impact of COVID-19 on the Socio-economic Achievement Gap Post - COVID Baseline PISA Scores in Reading Poorest % points in PISA scale ≈ year of schooling Richest % Note: About learning loss estimates: The World Bank has estimated the effects of COVID-19 related school closures on learning outcomes for 157 countries. Simulations use data on learning outcomes, years of schooling and monthly wages to estimate the potential effects of school closures in general and across socioeconomic groups. For global and regional estimates, see World Bank 2020a. In the medium term, the widening of the achievement gap and the income shock of the current pandemic may push up school dropouts, particularly for disadvantaged students. Students who fall behind in education will also be demotivated and at higher risk of dropping out. The loss in household incomes due to inability to work during COVID-19 will also test the ability of households to keep students in school, increasing out-of-school youth and hindering transition to post-secondary education. The effect of COVID-19 on education will have a decades-long impact on the economy unless authorities act to recover learning losses and protect the human capital of affected cohorts. Learning losses and reduced years of schooling for student cohorts affected by COVID-19 will reduce their expected earnings by an estimated 2.6 percent, (Author’s simulations based on PISA data and expected returns to education) which may amount to an overall economic loss of up to US$136 million (2011 PPP) every year. Kosovo needs to protect education spending, ensure remediation to recover learning losses, prevent student dropouts, and invest in building a resilient education system that can pivot to different forms of delivering learning in an effective manner, when faced with crises in the future. Schools in Kosovo lack sufficient digital resources to integrate distance learning and enable a resilient education system. About 70 percent of teachers in Kosovo have the necessary technical and pedagogical skills, time, and technical support to effectively integrate digital technology into the teaching and learning process (Figure 26). However, the schools are severely short of digital technology. Only about 15 percent of 15-year-old Kosovar students (PISA OECD 2018) attend schools where the school administration agrees that the number of digital devices available is sufficient for instruction, compared to the OECD average of about 60 percent. Similarly, only about 22 percent of students study in schools where sufficient digital devices are connected to the Internet, as compared to the OECD average of 65 percent. These shortages highlight the need to equip schools in Kosovo with better resources for assisted learning and to improve the quality of the teaching and learning process in the country. Impact of COVID-19 on the Education System 47 Figure 26. School Resources for Resilient Education Compared, Percent The school has su cient qualified 72.2 technical assistant sta . Teachers are provided with incentives to 56 integrate digital devices in their teaching An e ective online learning support platform 22 is available . E ective professional resources for teachers to 66 learn how to use digital devices are available . Teachers have su cient time to prepare 80.4 lessons integrating digital devices . Teachers have the necessary technical and pedagogical skills to integrate digital 72.2 devices in instruction . The availability of adequate software is 17.3 su cient Digital devices at the school are su ciently 20.7 powerful in terms of computing capacity The number of digital devices for 14.5 instruction is su cient . School's internet bandwidth or speed is 28.7 su cient . The number of digital devices connected 22.1 to the internet is su cient . Percentage of students Source: OECD 2018. 48 Impact of COVID-19 on the Education System Impact of COVID-19 on the Education System 49 07 Summary and Policy Recommendations Expanding access to preschool education and ensuring universal access to quality basic education is crucial to build a labor force that has the complete education and solid foundational skills needed to build a strong economy. Kosovo has made sound improvements in access to preprimary and primary education. It now needs to equitably expand coverage of quality preschool education and reverse the decline in lower and upper secondary enrollment. Increasing access to preschool education requires expansion of alternative financing models for preschool beyond public provision, providing incentives for private providers and hybrid public/private models, as well as ensuring minimum quality standards through licensing and training.32 At lower and upper secondary, it will be important to continue to support remedial and extra support programs delivered through NGOS and when possible to institutionalize them (e.g., learning centers) targeted especially to low income families, minority children, and rural students. Additionally, the student registry should be developed in order to better track academic performance over the student’s entire school career. This registry should also allow for an early warning system of at risk/low performers for better targeting of services that keep students in school. Reducing liquidity constraints of poor families through removing the tight exclusion criteria and improving the targeting of the social assistance scheme to cover more poor families with children would also support improvements in access to pre-university education. The Kosovo Education Strategic Plan lays out a roadmap with clear indicators for development of the education system. Impressively, it includes specific goals for children of the minority groups including Roma, Ashkali, and Egyptian communities. This roadmap can be further strengthened by including specific goals for regions—currently there are wide regional disparities in access to quality education. It is also necessary to equalize per-student financing across regions with additional funds for disadvantaged students or schools in remote locations. Kosovo’s spending on education as a percentage of GDP and as a share of government spending is similar to other middle-income countries, but the quality of education is very low. Concerted effort is required to increase quality at all levels of education. Kosovo needs a system of learning assessments to guide teacher instruction and motivate improvement over time, especially in lagging municipalities. The country would benefit from building on the strengths of a qualified and motivated teaching force by implementing a strong professional development program that caters to teacher needs and improves the quality of teaching and learning. The majority of schools across Kosovo lack adequate and good quality educational materials and infrastructure for a safe and conducive learning environment. Tackling this problem requires more efficient spending, starting with a review of the 5 options for improving the school financing formula (a recent 2021 study supported under KESIP33), selecting an option, and putting in place the policies and capacity building at the municipal level for its implementation. The wage bill in the country has limited fiscal space for much-needed quality-enhancing inputs for a successful teaching and learning process. The increase in the wage bill in the education sector accounted for most of the budget increase in the country which has almost tripled over the last decade. While wages have increased, the number of teachers has not decreased even though the number of students is declining. A careful stocktaking of the number of administrators, teachers, and specialists needed could inform studies at the municipal level on school network optimization. Given the fiscal space needed, municipalities in Kosovo should be supported to raise their own revenues to supplement the education grants they receive from the central government and empowered to improve the efficiency and effectiveness of their education spending. Empowering municipalities to work with community members to undertake school network optimization can also help to create much-needed fiscal space to invest in quality-enhancing inputs. There are a large number of schools operating with fewer than the legal minimum of students. However, school network optimization will require a detailed process with the necessary complete information on students, teachers, and educational institutions, their characteristics and capacities, and there must be buy-in from teachers and the local community. It will also require creative solutions like refurbishing extra classrooms to accommodate a preschool and even some new costs, like providing transportation in order to realize bigger savings. A first step is to update the school mapping report produced in 2016 and work with interested municipalities actively seeking to find savings that can be invested in school quality. 32 For a complete set of recommendations on ECD expansion, please see: Demas, Aliu, Coll-Black, Zafeirakou, Hankey, and Gotcheva. 2021. A Situational Analysis of Early Childhood Development Services in Kosovo. World Bank, Washington, DC. 33 Kosovo Education System Improvement Project, 2016-2021. 52 Summary and Policy Recommendations The poor quality of the education system is also limiting labor market outcomes and growth of economic activity in Kosovo. Most firms attempting to fill high to medium skill positions have encountered difficulties in finding people with the requisite skills. While the economy is demanding people with higher-level skills, the education system is struggling to equip students with even foundational skills as well as the socio-emotional and higher-order skills sought by the economy. COVID-19 is further testing the resilience of the education system in Kosovo and effort is required to recover learning losses and protect the human capital of affected cohorts. The situation presented during COVID-19 that forced the closing of schools and delivery of education remotely, illuminated the system’s lack of readiness and capacity to use information communication technology to assist learning. Almost all of Kosovo’s schools have access to broadband; the next step is to draft a comprehensive plan for digitizing education delivery and ensuring equity of connectivity, sufficient equipment, and skilled and supported teachers who are prepared to use technology to assist learning. Additionally, due to learning losses, it will be important to assess students and place them at the right level. Schools are recommended to use rapid standardized testing to identify students that have experienced the largest learning losses and assign them to the most appropriate compensatory or remedial program. Pedagogical models such as Teaching at the Right Level can be used to assign students to different groups after testing.34 As schools reopen, more than 60 percent of countries have indicated that they will use school-based assessment to track learning, while 30 percent plan to use national assessments for primary school, and 50 percent plan to use national assessment for secondary school.35 34 Teaching at the right level can be implemented via computer-assisted learning, through tutors, or tracking, as respectively shown in Muralidharan, Karthik, Abhijeet Singh, and Alejandro J. Ganimian. 2019. “Disrupting Education? Experimental Evidence on Technology-Aided Instruction in India.” American Economic Review, 109 (4): 1426-60. https://www.aeaweb.org/articles?id=10.1257/aer.20171112; Banerjee, Abhijit V., S. Cole, E. Duflo, and Leigh Linden. 2007. “Remedying Education: Evidence from Two Randomized Experiments in India.” The Quarterly Journal of Economics, 122 (3): 1235–64. https://academic.oup.com/qje/article-abstract/122/3/1235/ 1879525?redirectedFrom=fulltext, and Duflo, Esther, Pascaline Dupas, and Michael Kremer. 2011. “Peer Effects, Teacher Incentives, and the Impact of Tracking: Evidence from a Randomized Evaluation in Kenya.” American Economic Review, 101 (5): 1739-74. https://www.aeaweb.org/articles?id=10.1257/aer.101.5.1739 35 UNESCO, UNICEF and the World Bank (2020). Op. Cit. Summary and Policy Recommendations 53 Annex 1: Mapping of Gross Enrollment By Level of Education Annex 1A Figure 27: Mapping of Pre-school Enrollment in Kosovo Leposaviq . % . % 0.0% Mitrovicë e Veriut Zveçan 0.0% Zubin Potok Podujevë 0.0% 2.3% Istog Mitrovicë Prishtinë 12.6% 15.4% 34.0% Pejë Skënderaj Vushtrri Novobërdë 8.5% 3.0% 8.0% 0.0% Klinë Drenas Obiliq Kamenicë 7.4% 7.9% 10.0% 10.1% Deçan Malishevë Fushë Kosovë Gjilan 0.0% 3.0% 23.8% 14.2% Junik Suharekë Graçanicë Ranillug 16.1% 7.0% Rahovec Lipjan Kllokot 2.2% 7.0% 0.0% Gjakovë Shtime Ferizaj Partesh 11.1% 2.6% 13.3% Mamushë Shtërpcë Viti 0.0% 0.0% 6.1% Prizren Kaçanik 5.6% 3.3% Hani i Elezit 0.0% Dragash 0.4% Preschool enrollment is very low and it tends to be concentrated in more urban municipalities. 56 Summary and Policy Recommendations Annex 1B Figure 28: Mapping of Gross Enrollment at Pre-primary Education in Kosovo. Leposaviq . % . % 0.0% Mitrovicë e Veriut Zveçan 0.0% Zubin Potok Podujevë 13.2% 95.0% Istog Mitrovicë Prishtinë 73.0% 115.0% 108.0% Pejë Skënderaj Vushtrri Novobërdë 89.3% 85.5% 104.9% 41.7% Klinë Drenas Obiliq Kamenicë 98.3% 103.2% 104.2% 71.6% Deçan Malishevë Fushë Kosovë Gjilan 77.4% 111.4% 55.8% 115.0% Junik Suharekë Graçanicë Ranillug 72.2% 91.0% Rahovec Lipjan Kllokot 97.6% 112.4% 0.0% Gjakovë Shtime Ferizaj Partesh 75.2% 96.0% 107.3% Mamushë Shtërpcë Viti 79.9% 31.4% 78.9% Prizren Kaçanik 71.9% 92.6% Hani i Elezit 106.6% Dragash 57.2% Gross enrollment rates at pre-primary education are 92.5 percent as of 2019/20, but there are still some municipalities that are lagging behind such as Dragash and Kamenicë. It must be noted that while enrollment rates are nearly universal, the average daily attendance is only for two hours a day. Summary and Policy Recommendations 57 Annex 1C Figure 29: Mapping of Gross Enrollment Rates at Primary Education in Kosovo Leposaviq . % . % 1.1% Mitrovicë e Veriut Zveçan 7.8% Zubin Potok Podujevë 8.2% 108.3% Istog Mitrovicë Prishtinë 89.6% 115.6% 114.0% Pejë Skënderaj Vushtrri Novobërdë 94.9% 95.0% 114.4% 51.5% Klinë Drenas Obiliq Kamenicë 98.0% 103.9% 128.4% 74.6% Deçan Malishevë Fushë Kosovë Gjilan 67.2% 113.5% 144.1% 107.2% Junik Suharekë Graçanicë Ranillug 59.7% 98.3% Rahovec Lipjan Kllokot 91.8% 115.1% 31.3% Gjakovë Shtime Ferizaj Partesh 88.3% 105.5% 112.7% Mamushë Shtërpcë Viti 87.5% 43.6% 85.3% Prizren Kaçanik 84.9% 94.0% Hani i Elezit 99.5% Dragash 67.0% Primary enrollment rates are general close to universal, however, there are still municipalities like Decan, Dragash, and Junik that have far below universal coverage. 58 Summary and Policy Recommendations Annex 1D Figure 30: Mapping of Gross Enrollment at Lower Secondary Education in Kosovo. Leposaviq . % . % 0.8% Mitrovicë e Veriut Zveçan 4.1% Zubin Potok Podujevë 11.7% 103.1% Istog Mitrovicë Prishtinë 81.7% 101.6% 97.5% Pejë Skënderaj Vushtrri Novobërdë 83.4% 87.2% 106.9% 40.4% Klinë Drenas Obiliq Kamenicë 94.7% 89.6% 112.4% 73.6% Deçan Malishevë Fushë Kosovë Gjilan 61.3% 102.9% 104.8% 99.3% Junik Suharekë Graçanicë Ranillug 52.9% 88.8% Rahovec Lipjan Kllokot 81.4% 106.1% 28.3% Gjakovë Shtime Ferizaj Partesh 84.0% 98.5% 99.8% Mamushë Shtërpcë Viti 69.3% 43.6% 83.9% Prizren Kaçanik 75.3% 85.8% Hani i Elezit 90.6% Dragash 57.0% Wide variation exists in access to lower-secondary education with a number of municipalities in the western part of Kosovo falling behind. Summary and Policy Recommendations 59 Annex 1E Figure 31: Mapping of Gross Enrollment at Upper-secondary Education Leposaviq . % . % 0.0% Mitrovicë e Veriut Zveçan 0.0% Zubin Potok Podujevë 0.0% 91.8% Istog Mitrovicë Prishtinë 65.3% 110.1% 100.6% Pejë Skënderaj Vushtrri Novobërdë 83.0% 77.3% 82.4% 12.0% Klinë Drenas Obiliq Kamenicë 76.0% 78.1% 74.4% 67.4% Deçan Malishevë Fushë Kosovë Gjilan 48.8% 80.9% 72.9% 114.2% Junik Suharekë 52.5% 74.4% Graçanicë Ranillug Rahovec Lipjan Kllokot 55.6% 94.0% 0.0% Gjakovë Shtime Ferizaj Partesh 84.7% 86.0% 108.0% Mamushë Shtërpcë Viti 49.2% 38.1% 64.3% Prizren Kaçanik 68.6% 66.5% Hani i Elezit 55.7% Dragash 28.0% Source: Author’s calculations using data from Ministry of Education and Kosovo Agency of Statistics. Data reported for certain municipalities (primarily Serbian-majority) are incomplete. These include: Gracanice, Kllokot, Leposaviq, Mitrovice, Novoberdo, Ranillug, Pratesh, Sterpce, Zubin Potok and Zvecan. The gross enrollment rate at upper secondary education is 84 percent, but for some municipalities like Decan and Dragash, it is as low as 49 percent. 60 Summary and Policy Recommendations Summary and Policy Recommendations 61 Annex 2: Details of the Algorithm for School Network Optimization and Identification of Factors That May Impact A Simulation Designing an optimal network requires a very detailed set of information, not only about the school infrastructure but also about enrollment staffing, actual costs, as well as legal and political constraints at the local level. The simulation strategy depends on the number and tightness of constraints. In a typical municipality in Kosovo, the number of school units with available space is large enough to render the simulation and comparison of all theoretical possibilities too long to be practical. In that case, one needs to implement an empirical method to find a suitable solution.36 In some instances, however, where the number of schools with adequate available classrooms would be limited or where only a small number of schools would be close enough to consider a merger, the number of possibilities may be small enough to allow the comparison of all possibilities. The latter case is easier to solve while finding a suitable solution in the former require using meta-heuristic algorithm. Such an algorithm also helps to explore solutions outside of the local constraints, allowing identification of the impacts of each constraint on the possibilities of consolidation. The effects of distance and regulatory constraints on the possibilities of optimization can be explored using a dedicated algorithm. The algorithm main principle is to try fitting the smallest school units into the facilities with the most empty space. In details: i. One starts by ordering the school facilities in the municipality by the number of empty classrooms. Depending on the availability of information, this exercise can be restrained to the number of empty classrooms in adequate condition. One can also exclude from the set of facilities the ones which do not fulfill a list of minimum criteria. ii. One can symmetrically order the school units by the number of groups (classes), assuming a certain maximum class size, in accordance with the regulation. iii. One then can select the school facility with the most room available as the receiving school, which is merged with the smallest school unit (i.e. the ones with the smallest number of groups/classes) into the school facility (receiver school) with the most room available. iv. Distance conditions are then checked, and this merger is only considered if the distance between the two school units falls below a given threshold, which can be different depending on the condition (for instance for school units above a certain altitude). v. Then one checks that after the merger, the new number of groups in the consolidated school unit fits within the facility of the receiving school. vi. Steps one to five are reiterated until no school unit can fit. Principle of the school network optimization algorithm School units are School facilities are ordered by size ordered by # of empty classrooms Random Draw Receiving school facilities of similar vacant capacity are also jumbled Small school units are integrated into empty school facilities until there is no school unit large enough to receive it Source: Author’s illustration 36 Which may not be the absolute best one since one would need to run through all the possibilities to identify it. 64 Summary and Policy Recommendations The school mapping data from 2016 and enrollment and staff numbers in 2019 cannot be matched automatically. Several practical issues are preventing a perfect matching. First, the status of some facilities in the school mapping database of 2016 may have been already changed including closure or new school units may have been merged or opened since then. An update of the school mapping and alignment of the current teacher and student data from the EMIS is required. Second, schools are not identified with a common identifier in the school mapping database and in the student enrollment file which comes from the EMIS. Some school names may differ because of changes. Third, when a school institution has more than one satellite, it is not always possible to pair each satellite with a given facility, especially when the satellites are in the same locality. In the student enrollment database, satellite schools are identified as a combination of the name of the school institution and an integer, while school facilities are rather identified by an address. As a consequence, the pairing of school units and school facilities needs to be completed manually, which takes time. Pairing data about facilities and enrollment by grade at school units allows estimating the number of groups and empty classrooms. The theoretical number of groups can be easily estimated from the maximum number of students allowed per class, assuming that no more than two distinct grades can be taught in the same group. The number of empty classrooms is the number of classrooms in the facilities minus the number of groups computed as indicated above. Summary and Policy Recommendations 65 References “1302.21 Center-Based Option. | ECLKC.” n.d. Accessed June 4, 2020. https://eclkc.ohs.acf.hhs.gov/policy/45-cfr-chap-xiii/1302-21-center-based-option. Autor, D., F. Levy and R. J, Murnane, 2003, “The Skill Content of Recent Technological Change: An Empirical Exploration,” The Quarterly Journal of Economics, vol. 118(4), pp 1279-1333. Construction Management and Consulting. 2016. School Map Report, National Level, 2016. Inventory of Educational Facilities in Kosovo. World Bank, Kosovo. Demas, Angela, Mrike Aliu, Sarah Coll-Black, Aglaia Zafeirakou, Aline Hankey, and Boryana Gotcheva. 2021. A Situational Analysis of Early Childhood Development Services in Kosovo. World Bank, Washington, DC. Dunbar, M. 2013. Engaging the private sector in skills development, Health & Education Advice and Resource Team, https://www.educationinnovations.org/European Union. 2019. “Mid Term Evaluation: Implementation of Kosovo Education Strategic Plan” International Association for the Evaluation of Educational Achievement (IEA). 2019. Trends in International Mathematics and Science Study (TIMSS) 2019, International Results in Mathematics and Science. Downloaded from http://timss2019.org/download Kosovo Agency of Statistics and UNICEF. 2020. Kosovo Multiple Indicator Cluster Survey 2019– 2020 and Roma, Ashkali and Egyptian Communities Multiple Indicator Cluster Survey 2019-2020, Survey Findings Report. Prishtina, Kosovo: Kosovo Agency of Statistics and UNICEF. Kosovo Education and Employment Network – KEEN. November 2018. Pre-school Education in Kosovo. Prishtina. Ministry of Education, Science, Technology and Innovation (MESTI) and the Kosovo Agency of Statistics (KAS). 2020. Education Statistics in Kosovo 2019/2020. Prishtina, Kosovo. MESTI. 2016. Kosovo Education Strategic Plan, 2016-2021. Government of Kosovo, Prishtina. European Commission. June 2019. Kosovo plans expanded accesses to early childhood education and care. ESPN Flash Report 2019/30. European Commission. OECD. 2020. Education spending (indicator). doi: 10.1787/ca274bac-en (Accessed on 13 May 2020). OECD. PISA 2018 Saqipi, Blerim. 2018. Modernizing school system in Kosovo through sustainable and coherent teacher development. Saqipi, Blerim., Angela Demas, Lindsay Adams. 2021. SABER – Early Childhood Development Kosovo Country Report. World Bank, Washington, DC. UNICEF. 2020. Multiple Indicator Cluster Surveys, Kosovo 2019-2020. World Bank. 2020a. Simulating the Potential Impacts of COVID-19 School Closures on Schooling and Learning Outcomes: A Set of Global Estimates (English).  Washington, D.C.: World Bank Group. http://documents.worldbank.org/curated/en/885601592484683223/Simulating-the-Potential- Impacts-of-COVID-19-School-Closures-on-Schooling-and-Learning-Outcomes-A-Set-of-Global- Estimates World Bank. 2020b. “The Economic and Social Impact of COVID – 19.” Western Balkans Regular Economic Report No.17, Spring 2020. World Bank. 2018a. “Kosovo Country Report: Findings from the Skills Towards Employment and Productivity Survey.” World Bank, Washington DC. World Bank. 2018b. World Development Report 2018: Learning to Realize Education’s Promise. Washington, DC: World Bank. doi:10.1596/978-1-4648-1096-1. License: Creative Commons Attribution CC BY 3.0 IGO. 66 WORLD BANK GROUP Enhancing Human Capital by Improving Education Kosovo Country 68 November 2021 Summary and Policy Recommendations Economic Memorandum