TWO YEARS AFTER SAVING A GENERATION Two Years After | Sa- ving a Generation IN COLLABORATION WITH Two Years After | Saving a Generation © 2022 International Bank for Reconstruction and De- velopment/The World Bank Group 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 Group with external contributions. The findings, in- terpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank Group, its Board of Executive Directors, or the govern- ments they represent. The World Bank Group does not guarantee the accura- cy, completeness, or currency of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. 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 Group concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be construed or con- sidered to be a limitation upon or waiver of the privi- leges and immunities of The World Bank Group, all of which are specifically reserved. Rights and Permissions The material in this work is subject to copyright. Be- cause The World Bank Group encourages dissemina- tion 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 subsidi- ary rights, should be addressed to World Bank Publica- tions, The World Bank Group, 1818 H Street NW, Wash- PHOTO BY: © THE WORLD BANK ington, DC 20433, USA; fax: 202-522- 2625; e-mail: pubrights@worldbank.org. Cover photo: Illustration based on photographs by: Ljupco Smokovski/Shutterstock and New Africa/Shut- terstock. Further permission required for reuse. Design: Alejandro Espinosa/sonideas Index Two Years After | Saving a Generation 3 Index Acknowledgements 11 Foreword 13 List of acronyms 14 Executive summary 16 Introduction 26 CHAPTER 1 The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response 28 1.1 A health and economic crisis like no other…disproportionately affecting LAC 29 1.2 The COVID-19 pandemic and its impact on LAC’s education systems to date 31 1.3 Overview of the remote learning response to the pandemic 35 1.3.1 The elements of a complex remote learning response 35 1.3.2 The limitations of the remote learning response 38 1.4 Key takeaway messages 45 CHAPTER 2 The impact, so far, of the COVID-19 pandemic on schooling 46 2.1 The state of schooling prior to the pandemic: A positive outlook with good prospects 47 2.2 The COVID-19 pandemic shock and its impact on schooling 50 2.3 Key takeaway messages 58 CHAPTER 3 The impact of the COVID-19 pandemic on learning outcomes and skills development 60 3.1 The impact on learning 61 3.1.1 The state of learning achievement in LAC prior to the pandemic: A grim picture 61 3.1.2 The COVID-19 pandemic shock and its impact on learning: A crisis within a crisis 65 3.2 The impact on transferable skills 79 3.2.1 Transferable skills in LAC prior to the pandemic 79 3.2.2 The impact of the pandemic on transferable skills 82 Index 4 Two Years After | Saving a Generation 3.3 The implications of the pandemic for digital skills 82 3.3.1 Digital skills and “ICT preparedness” prior to the pandemic 82 3.3.2 Effects of the pandemic on digital skills 84 3.4 The impact on psychosocial health and well-being 85 3.5 Key takeaway messages 86 CHAPTER 4 Recovering and accelerating: Returning to schooling 88 4.1 Reopening schools 90 4.2 Re-enrolling and preventing dropouts 94 4.3 Key takeaway messages 100 CHAPTER 5 Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 102 5.1 Recovering from learning losses 103 5.1.1 Prioritizing the curriculum 104 5.1.2 Assessing learning levels and losses 109 5.1.3 Implementing learning recovery programs 116 5.2 Addressing psychosocial health and well-being 121 5.3 Addressing digital gaps 125 5.4 Key takeaway messages 129 CHAPTER 6 Conclusion 132 References 136 Annexes 144 Figures Figure ES.1. Contribution to global totals for population and number of COVID-19 related deaths, by region (%), March 2020 – March 2022 18 Figure ES.2. Number of school weeks with schools fully and partially closed, by region, March 2020 – March 2022 18 Figure ES.3. Internet access across LAC, selected countries (%), latest available year 19 Figure ES.4. Attendance rates before and during the pandemic, selected countries (%), February 2020 vs. May/July 2021 19 Figure ES.5. TERCE 2013, ERCE 2019 and simulated 2022 reading scores 20 Figure ES.6. TERCE 2013, ERCE 2019 and simulated 2022 math scores 20 Figure ES.7. The road to recovery and acceleration in LAC 21 Figure ES.8. The four commitments of the education recovery in LAC 22 Figure ES.9. Examples of learning remediation and recovery programs in LAC 24 Index Two Years After | Saving a Generation 5 Figure 1.1. Total cumulative number of deaths per million people, by region, 2020 - 2022 29 Figure 1.2. Contribution to global totals for population and number of COVID-19 related deaths, by region (%), 2020-2022 30 Figure 1.3. Share of population fully and partially vaccinated, by region (%), 2020-2022 30 Figure 1.4. Index of real GDP growth rate, by region, 2018-2023 (est.) 31 Figure 1.5. Increase in the unemployment rate during the pandemic, by region (%), 2019-2020 and 2019-2021 31 Figure 1.6. School calendars in LAC and timing of the COVID-19 pandemic: Understanding the implications 32 Figure 1.7. Number of school weeks with schools fully and partially closed, by region, March 2020 - March 2022 33 Figure 1.8. School closure intensity (%) by country, March 2020 - March 2022 33 Figure 1.9. Dynamics of school closures and reopening in LAC countries, by end of quarter (% of total school weeks by status), March 2020 – March 2022 34 Figure 1.10. Summary of school closures for LAC, by country (share of school weeks with schools fully or mostly closed to date) 35 Figure 1.11. School closure index: Top 15 LAC countries 35 Figure 1.12. Average number of instruction days lost during 2020, by level of education and region 36 Figure 1.13. Channels implemented for learning continuity in LAC during 2020 (% of countries) 36 Figure 1.14. Support provided to teachers nationwide during 2020, by type of support and region (%) 37 Figure 1.15. Number of interactions encouraged between teachers and parents and/or students during school closures in 2020, by region (%) 38 Figure 1.16. Internet access across LAC, selected countries (%), latest year available 40 Figure 1.17. Households in LAC that reported having problems accessing the Internet (%), mid-2021 40 Figure 1.18a. Share of primary students with a computer in the household by ethnic origin, selected countries (%) 41 Figure 1.18b. Share of primary students with Internet in the household by ethnic origin, selected countries (%) 41 Figure 1.18c. Share of secondary students with computer in the household by ethnic origin, selected countries (%) 41 Figure 1.18d. Share of secondary students with Internet in the household by ethnic origin, selected countries (%) 41 Figure 1.19. Remote learning implementation strategy in LAC during 2020, selected countries 42 Figure 1.20. Disengagement rate, by country (% of student-age population), 2020 vs 2021 43 Figure B.1.2.1. Dominican Republic: Proportion of students that watched distance learning programs in Gran Santo Domingo and Santiago (% of students aged 4-17) 44 Figure 2.1. Enrollment vs. school-age population in LAC (million), 1989-2019 47 Figure 2.2. Population growth rate in LAC (%), 1970 - 2019 47 Figure 2.3. Death rate index vs. birth rate index in LAC (per 1,000 people), 1970-2019 48 Figure 2.4. School-age population index, LAC (total and by education level), 1970-2019 48 Figure 2.5. Total enrollment in LAC, by level of education (million), 2000-2019 49 Figure 2.6. Net enrollment rate in LAC, by level of education (%), 2000-2019 49 Figure 2.7. Evolution of enrollment in LAC for 4 cohorts, 1990-2010 (% enrolled) 50 Figure 2.8. Distribution of upper secondary education (USE)-age students in LAC by schooling status for 4 cohorts entering USE between 1994 and 2011 (%) 50 Figure 2.9. OOSP in LAC, by compulsory education level and for selected years (million), 2000-2019 51 Figure 2.10. OOSP in LAC, as a share of school-aged population, by compulsory education level and for selected years (%), 2000-2019 51 Figure 2.11. OOSP in LAC, by gender and compulsory education level (thousands), 2019 51 Index 6 Two Years After | Saving a Generation Figure 2.12. OOSP in LAC, as a share of school-aged population, by gender and compulsory education level (%), 2019 51 Figure 2.13. Possible trajectories of the OOSP in LAC after the COVID-19 pandemic (million) 52 Figure 2.14. Students at risk of dropping out in LAC, by education level (thousands and %) 52 Figure 2.15. Dominican Republic: Total enrollment index, by level of education, 2019-2021 53 Figure 2.16. Panama: Total enrollment index, by school age group, 2019-2021 53 Figure 2.17. Attendance rates before and during the pandemic, for selected countries (%), February 2020 vs. May/July 2021 54 Figure 2.18. Attendance rates before and during the pandemic, by age group for selected countries (% change), February 2020 vs. May/July 2021 54 Figure 2.19. Argentina: OOSP as a share of the corresponding school-aged population, total and by age group (%), 2019-2021 55 Figure 2.20. Brazil: OOSP as a share of the corresponding school-aged population, total and by age group (%), 2019-2021 55 Figure 2.21. Colombia: OOSP as a share of the corresponding school-aged population, total and by age group (%), 2019-2021 55 Figure 2.22. Costa Rica: OOSP as a share of the corresponding school-aged population, total and by age group (%), 2019-2021 55 Figure B.2.1.1. Dominican Republic: Enrollment, 2019-2021 57 Figure B.2.1.2. Guatemala: Enrollment, 2019-2021 57 Figure B.2.1.3. Panama: Enrollment, 2019-2021 57 Figure 2.23. LAC: Attendance rates before and during the pandemic, by age group and access to assets (% difference) 59 Figure 2.24. LAC: Gender gaps in attendance rates before and during the pandemic, by access to assets (%) 59 Figure 3.1a. Share of students in grade 3 below MPL in reading (%), ERCE 2019 62 Figure 3.1b. Share of students in grade 3 below MPL in math (%), ERCE 2019 62 Figure 3.1c. Share of students in grade 6 below MPL in reading (%), ERCE 2019 62 Figure 3.1d. Share of students in grade 6 below MPL in math (%), ERCE 2019 62 Figure 3.2a. Learning gains in reading and math for students in grade 3, TERCE 2013 and ERCE 2019 63 Figure 3.2b. Learning gains in reading and math for students in grade 6, TERCE 2013 and ERCE 2019 63 Figure 3.3a. Student achievement differences by gender in Reading, 3rd and 6th grade, ERCE 2019 64 Figure 3.3b. Student achievement differences by gender in Math, 3rd and 6th grade, ERCE 2019 64 Figure 3.4. Factors most strongly associated with student performance in 6th grade, ERCE 2019 65 Figure 3.5. Learning trajectories pre- and post-COVID-19, and implications of current learning losses on future learning 66 Figure 3.6. Simulated loss in learning-adjusted years of schooling (LAYS) due to COVID-19 in LAC 68 Figure 3.7. Simulated learning loss effects on mean ERCE 2019 scores, by grade and subject in LAC (%) 68 Figure 3.8. Simulated learning loss effects on the distribution of ERCE 2019 proficiency levels, by grade and subject in LAC (%) 69 Figure B.3.1.1. Learning poverty, globally and by region (%), 2015 – 2022 (est.)* 70 Figure 3.9. Simulated learning loss effects on the share of students below MPL in ERCE 2019, by grade and subject in LAC (%) 71 Figure 3.10. Colombia: Share of Grade 3 students at each reading level by frequency of attendance (%), 2020 72 Figure 3.11. Colombia: Share of Grade 3 students at each math level by frequency of attendance (%), 2020 72 Index Two Years After | Saving a Generation 7 Figure B.3.2.1. Colombia: Average SABER-11 test results, 2015-2021 73 Figure B.3.2.2. Colombia: Average SABER-11 test results by SES, 2019-2021 73 Figure 3.12. Brazil: National SAEB score results in language: Time trend by grade in São Paulo, 2011-2021 75 Figure 3.13. Brazil: National SAEB score results in mathematics: Time trend by grade in São Paulo, 2011-2021 75 Figure B.3.3.1. Guanajuato: RIMA average scores by level, grade, and subject, 2020 vs. 2021 76 Figure 3.14. Mexico: Learning loss in Mathematics, by SES level, 2019 vs. 2021 78 Figure 3.15. Brazil: Share of 6/7-years-old children who cannot read and write, by income level (%), 2019 vs. 2021 79 Figure 3.16. Core transferable skills and dimensions 80 Figure 3.17. PISA 2018 index of student cooperation for LAC countries 81 Figure 3.18. Chile: Share of students reporting socio-emotional learning at school, by grade (%), 2020 82 Figure 3.19. Share of students in LAC countries whose school principals agreed with the statement “Teachers have the necessary technical and pedagogical skills to integrate digital devices in instruction” (%), 2018 83 Figure 3.20. Share of teachers by level of digital skills, by region (%), 2020 84 Figure 3.21. Examples of digital toolkits with teaching resources 85 Figure 3.22. Argentina: Share of youth reporting feeling distressed, scared, or depressed by the uncertainty of the pandemic (%), April 2020 – October 2021 86 Figure 4.1. The road to recovery and acceleration in LAC 89 Figure 4.2. Some critical safety standards for reopening schools 91 Figure 4.3. Vaccination of the population and school reopening rates in LAC 91 Figure 4.4. Link between early shock before age six and well-being later in life 93 Figure 4.5. Successful policies to spur enrollment and attendance, and deter dropouts 97 Figure 4.6. Countries in LAC with previous experiences with cash transfers 97 Figure 4.7. Chile: Early Warning System 99 Figure 4.8. Brazil: Early Warning System 99 Figure 5.1. Three-dimensional model of curriculum, pedagogy, and assessment 103 Figure 5.2. Number of LAC countries that prioritized certain areas of the curriculum or certain skills due to COVID-19, by year and school cycle, 2021 105 Figure 5.3. Programmatic intervention areas through multiple pathways 105 Figure B.5.1.1. Key actions for refocusing school management on learning outcomes and foundational skills 107 Figure B.5.1.2. Teaching at the right level (TaRL) 108 Figure 5.4. Map of countries based on decisions made regarding the administration of learning assessments in 2020 and 2021 110 Figure 5.5. Continuity, changes, and innovations as a reaction to the pandemic in national assessments for monitoring purposes and examinations to certify and select individual students 111 Figure 5.6. Number of LAC countries that adjusted certification examinations due to the pandemic during school year 2019/20 and 2020/20, by type of adjustment 112 Figure 5.7. Different types of learning recovery programs in LAC 116 Figure 5.8. Number of LAC countries that implemented wide remedial measures to address learning gaps when schools reopened after the first closure in 2020, by type of remedial measure 118 Figure 5.9. Key questions for countries to consider when implementing and scaling-up learning recovery programs 119 Index 8 Two Years After | Saving a Generation Figure B.5.5.1. Pre-primary students in LAC: Remote learning during school closures, and support received following reopening, 2021 123 Figure B.5.6.1. Evaluating key recovery policies for 2022 126 Figure 6.1. Allocation to education in total COVID-19 stimulus packages, by region (million USD and %), 2021 133 Figure 6.2. The four commitments of the education recovery in LAC 135 Boxes Box 1.1. The remote learning experience in Chile 39 Box 1.2. Assessing the effectiveness of remote learning in the Dominican Republic during school closures 44 Box 2.1. Dropouts in private but not in public schools: A push-and-pull effect? 57 Box 3.1. Learning poverty in LAC: A serious issue, deepening with COVID-19 70 Box 3.2. Assessing learning in difficult times of the pandemic: Colombia’s culture of evaluation 73 Box 3.3. Measuring learning and promoting alliances for learning recovery: The case of Guanajuato 76 Box 4.1. Challenges for and support to school reopening in Peru 92 Box 4.2. School reopening in federal countries: Challenges and lessons learned 95 Box 4.3. Consolidating more robust hybrid models: An opportunity that cannot be missed 96 Box 5.1. Accelerating: Foundational skills for the future 106 Box 5.2. PLaNEA’s Program: An innovative approach to develop transferable skills 108 Box 5.3. Examples of formative assessments designed at the central level. Brazil, State of Ceará: Diagnostic Evaluations 115 Box 5.4. Example of a learning recovery program: Mexico. Teach at the Right Level (TARL)/ Medición Independiente de Aprendizajes (Independent Learning Assessment) (MIA) 117 Box 5.5. Recovering and accelerating in ECE 122 Box 5.6. The strengths of the recovery response in Chile 126 Box 5.7. Tools and surveys for teachers’ well-being in Peru 127 Box 5.8. The Long-term vision of the role of education technologies in accelerating learning and reducing learning gaps in Panama 130 Box A.5.2. Digital competencies frameworks and assessments: A priority for post-pandemic times 152 Tables Table 3.1. Selected studies for LAC on learning losses 77 Table 3.2. Unequal learning losses: Increase in learning losses, by vulnerability characteristic 78 Table 5.1. Examples of centrally designed formative assessments in LAC 114 Table 5.2. Critical features needed to adapt, scale-up, and ensure sustainability of flagship learning recovery programs 120 Table 5.3. Selected initiatives to improve Internet connectivity for schools and students in LAC 128 Two Years After | Saving a Generation 10 Two Years After | Saving a Generation PHOTO BY: © MACIEJ CZEKAJEWSKI / SHUTTERSTOCK.COM Acknowledgements Two Years After | Saving a Generation 11 Acknowledgements T  is document was prepared by a core team from the Education Units of the Latin America and the Caribbean h Region of the World Bank Group (HLCED) and UNICEF (LACRO), and also benefitted with the collaboration from UNESCO’s Latin American Laboratory for Assessment of the Quality of Education (LLECE). The drafting of the report was co-led by Emanuela di Gropello (Practice Manager, HLCED), who also provided overall guidance to the team, and Juan Diego Alonso (Senior Economist, HLCED). Core contributors to the book included, in addition: (i) the following staff of and consultants to HLCED: Horacio Álvarez Marinelli, Diego Ángel Urdinola, Juliana Chen Peraza, Wendy De León Samayoa, Verónica Díaz Hinostroza, Ildo Lautharte, María José Ramírez, Helena Rovner, Erika Schutt Pardo, Denise Stolt, María José Vargas, and Mónica Yáñez Pagans; (ii) the following staff of and consultants to LACRO: Ruth Custode, Vincenzo Placco, María Paula Reinbold, Claudio Santibáñez Servat, and Liora Schwartz; and iii) the following staff of and consultants to LLECE: Carlos Cayumán Cofré, Francisco Gatica Eguiguren, and Carlos Henríquez Calderón.  he document also benefitted from the overall guidance, advice and insights from Jaime Saavedra (Global Di- T rector, Education Global Practice, Human Development Vice Presidency, World Bank Group), Luis Benveniste (Regional Human Development Director, Latin American and the Caribbean Vice Presidency, World Bank Group), Italo Dutra (Regional Education Advisor, UNICEF LACRO), and Claudia Uribe Salazar (Director of Regional Bureau and Representative in Chile, UNESCO). Peer review contributions which greatly enriched the report were provid- ed by Halsey Rogers (Lead Economist, Education Global Practice, Human Development Vice Presidency, World Bank Group), Cristóbal Cobo (Senior Education Specialist, Education Global Practice, Human Development Vice Presidency, World Bank Group) and Ciro Avitabile (Senior Economist, HLCED). Additional valuable comments and contributions were made by Gabriel Demombynes (Human Development Program Leader for Colombia, Hu- man Development Vice Presidency, World Bank Group), Rita Almeida (Human Development Program Leader for Central America and the Dominican Republic, Human Development Vice Presidency, World Bank Group), Joao Pedro Wagner de Azevedo (Lead Economist, Education Global Practice, Human Development Vice Presidency, World Bank Group), Marie-Hélène Cloutier (Senior Economist, Education Global Practice, Human Development Vice Presidency, World Bank Group), Daniela Trucco (Senior Social Affairs Officer, Economic Commission for the Latin America and the Caribbean), Mariana Huepe (Social Affairs Officer, Economic Commission for the Latin America and the Caribbean), Yi-Ning Wong (Junior Professional Associate, Education Global Practice, Human Development Vice Presidency, World Bank Group), and Maryam Akmal (Young Professional, Education Global Practice, Human Development Vice Presidency, World Bank Group).  everal other members of the HLCED team of the World Bank Group, of UNICEF LACRO, and UNESCO’s LLECE also S provided specific inputs, and/or participated in key discussions around the report, which builds on many coun- try-specific analytical and policy diagnostics. A special mention is given to the Inter-American Dialogue (IAD), which commissioned two strategic background papers that helped shape and deepen the analysis for several aspects of chapter 5 and which also helped organize the “Recovering Schooling and Learning After Covid-19 in LAC” series of webinars that greatly enriched discussions around key topics featured in the book. We would also like to particularly thank the Ministries of Education of Chile and Colombia and the Learning Assessment Agen- cies of each of these two countries, for kindly facilitating dedicated inputs that were key to the report. Finally, we would like to thank Paula Flores Carrillo for exceptional administrative support throughout the planning and production process. 12 Two Years After | Saving a Generation FOTOGRAFÍA DE: © J.P. JUNIOR PEREIRA / SHUTTERSTOCK.COM Foreword Two Years After | Saving a Generation 13 Foreword I  the last two years, the COVID-19 pandemic has wreaked havoc in Latin America and the Caribbean. The re- n gion has suffered a triple curse, as it faced the largest combined impact in health, economic and educational terms. The disproportionate impact of the pandemic on people´s lives, livelihoods, and human capital formation represents, without doubt, one of the worst crises in LAC’s history. As we seek to rebuild better and foster more inclusive and sustainable growth, the main concern, nonetheless, is not the heavy toll of the pandemic, but the future of an entire generation of children and young people who have endured this massive shock.  rior to the pandemic, LAC was well-known to be one of the most unequal regions in the world. Disparities were P notable not just in terms of income but, more importantly, in terms of opportunities. Education had been the great equalizer, steadily helping build a fairer society. But it simply vanished with the sudden closure of schools across the region and prolonged suspension of in-person learning.  n other words, the outbreak of the coronavirus widened existing gaps in access to high-quality education, teach- I ers, and schools between students with resources – those with access to Internet, a computer at home, and a school that could quickly implement virtual education through online platforms – and more vulnerable students – those that did not have access to Internet, nor a computer at home or a school that could deliver virtual learn- ing. This brought about a new segregation pattern, leading to a crisis within a crisis.  his report is the first evidence-based assessment of this educational catastrophe in Latin America and the T Caribbean. The report intends to systematically document the impact that COVID-19 has had on the region’s education sector two years after. The 24 months since the outbreak of the pandemic in March 2020 is described sequentially, focusing firstly on the features of the “triple curse”, and then on the direct impact on schooling, learning and skills development. The report also addresses significant cross-sectoral impacts, namely those re- lated to digital and transferable skills.  n addition to assessing the impact, we must find ways to improve outcomes and promote a better future for all. I The main goal of this report is to raise the alarm that we must all act jointly and urgently…to save a generation. A generation that has suffered like no other and that depends on our commitment to the education recovery agen- da now. The report sets out a clear and simple dual strategy, focused on the return to schooling and the learning recovery. We can still revert learning losses and, most importantly, seize the moment to build better and more inclusive and resilient education systems.  We can invest and act now or leave an entire generation behind. The choice is ours. Carlos Felipe Jaramillo, Vice President, Latin America and the Caribbean, World Bank Jean Gough, Regional Director for Latin America and the Caribbean, UNICEF List of acronyms 14 Two Years After | Saving a Generation List of acronyms CABA Ciudad Autónoma de Buenos Aires (Autonomous ICFES Instituto Colombiano para la Evaluación de la City of Buenos Aires) Calidad de la Educación (Colombian Institute for Education Quality Assessment) CCT Conditional Cash Transfer ICILS International Computer and Information CXC Caribbean Examinations Council Literacy Study DIA Diagnóstico Integral de Aprendizajes ICT Information and Communications Technologies (Comprehensive Learning Diagnostics) IPNV In-Pandemic, no Vaccination Available DR Dominican Republic IPVA In-Pandemic, Vaccination Available EAP East Asia and the Pacific LAC Latin America and the Caribbean ECA Europe and Central Asia LAYS Learning-Adjusted Years of Schooling ECE Early Childhood Education LFR Leadership for Renewal EGMA Early Grade Mathematics Assessment LLECE Laboratorio Latinoamericano de Evaluación de EGRA Early Grade Reading Assessment la Calidad de la Educación (The Latin American Laboratory for Assessment of the Quality of EMIS Education Management Information System Education) ENDO Encuesta Nacional Docente (National Survey of LMIC Lower Middle-Income Country Teachers) LSE Lower Secondary Education ERCE Estudio Regional Comparativo y Explicativo (Regional Comparative and Explanatory MENA Middle East and North Africa Study) MIA Medición Independiente de Aprendizaje EWS Early Warning System (Independent Learning Assessment) GDP Gross Domestic Product MPL Minimum Proficiency Level GSL Global School Leaders NA North America GTO Guanajuato (State of Mexico) OECD Organisation for Economic Co-operation and Development HCR Human Capital Review OOSP Out-of-School Population HFPS High-Frequency Phone Survey PERCE Primer Estudio Regional Comparativo y IAD Inter-American Dialogue Explicativo (First Regional Comparative and Explanatory Study) ICCS International Civic and Citizenship Education Study PIRLS Progress in International Reading Literacy Study List of acronyms Two Years After | Saving a Generation 15 PISA Programme for International Student SES Socio-Economic Status Assessment SMS Short-Message Service PNAD-C Pesquisa Nacional por Amostra de Domicílios Contínua (National Continuous Household SoR Science of Reading Survey) SSES Survey on Social and Emotional Skills PP Pre-Pandemic TaRL Teaching at the Right Level PPP Purchasing Power Parity TERCE Tercer Estudio Regional Comparativo y RIMA Recopilación de Información para la Mejora de Explicativo (Third Regional Comparative and los Aprendizajes (Data Collection for Improving Explanatory Study) Learning Outcomes) TIMSS Trends in International Mathematics and RtI Response to Intervention Science Study SA South Asia TV Television SSA Sub-Saharan Africa UNDP United Nations Development Program SAEB Sistema de Avaliação da Educação Básica (Basic UNESCO United Nations Educational, Scientific and Education Assessment System) Cultural Organization SDG Sustainable Development Goal UNICEF United Nations Children’s Fund SEG Secretaría de Educación de Guanajuato USE Upper Secondary Education (Ministry of Education of the State of Guanajuato) WBG World Bank Group SERCE Segundo Estudio Regional Comparativo y WHO World Health Organization Explicativo (Second Regional Comparative and Explanatory Study) 16 Two Years After | Saving a Generation PHOTO BY: © CARLOS MACIAS ROMO / SHUTTERSTOCK.COM Executive summary Executive summary Two Years After | Saving a Generation 17 Key Takeaways • LAC has endured one of the longest spells of school closures. The region was hit disproportionately hard in health, economic, and educational terms. In the region, an entire generation of students – approximately 170 million – were fully deprived of in-person education for roughly 1 out of 2 effective school days to date. • The effects of the pandemic on the education sector of the region have been severe: •  The overall impact on the out-of-school population was limited by the end of 2021 thanks to protective poli- cies, but more data are needed to fully assess this impact in 2021/2022 as millions of children and teenagers are at risk of dropping out for falling behind academically. •  Expected and real learning losses are very high, and more severe for earlier grades, younger children, and chil- dren from lower socio-economic backgrounds. Average primary education scores in reading and math would fall to levels of more than 10 years ago, in a context where improvements were already very slow. About 4 in 5 sixth graders may not be able to adequately understand and interpret a text of moderate length. •  Learning losses would translate into a decrease of about 12 percent in lifetime earnings for a student at school today. •  Psychosocial health and well-being have also been greatly affected. • The incipient recovery must focus on returning to schooling and, especially, recovering and accelerating learning: •  The return to schooling agenda should encompass: (i) safely and sustainably reopening all schools; (ii) re-en- rolling all students; and (iii) preventing dropouts. •  The recovering and accelerating learning agenda must comprise: (i) prioritizing and consolidating curricula; (ii) assessing learning levels; and (iii) implementing at scale learning recovery strategies and programs. • In a nutshell, this agenda entails the urgent and comprehensive implementation of four commitments: •  A commitment to place the education recovery at the top of the public agenda. •  A commitment to reintegrate all the children that abandoned school and ensure they stay in it. •  A commitment to recover lost learning and ensure the socio-emotional well-being of children. •  A commitment to value, support and train teachers. Executive summary 18 Two Years After | Saving a Generation Figure ES.1. Contribution to global totals for Figure ES.2. Number of school weeks with schools population and number of COVID-19 related deaths, by fully and partially closed, by region, region (%), March 2020 – March 2022 March 2020 – March 2022 Sub-Saharan Africa North America 7 56 East Asia and the Paci c Europe and Central Asia 13 15 East Asia and the Paci c 16 17 South Asia Sub-Saharan Africa 18 13 Middle East and North Africa Middle East 24 21 Europe and Central Asia and North Africa Latin America Latin America and 28% 29 29 and the Caribbean the Caribbean 8% South Asia 35 31 North America 0 10 20 30 40 50 60 70 0% 10% 20% 30% 40% Avg. weeks Avg. weeks Contribution to Deaths Contribution to Population of full closures of partial closures Source: Own elaboration on the basis of information from WHO COVID-19 Source: Own elaboration on the basis of UNESCO’s Global Monitoring of Dashboard. Updated as of March 31, 2022. School Closures. Updated as of March. 31, 2022. M ore than two years have passed since the COVID-19 pandemic hit the Latin America and the Caribbean (LAC) region leaving a gigantic trail of scars and wounds While the pandemic took a toll on education systems across the LAC region, notable efforts were made to continue delivering education remotely while schools along the way. LAC was hit disproportionately harder in were closed. To ensure teaching continuity, governments health, economic, and educational terms. The region has across the region were forced to combine a series of dis- experienced, on average, more COVID-19 deaths per cap- tance learning strategies. Most of the countries (or sub- ita than any other region in the world. In fact, while being national entities in federal countries where education home to only 8 percent of the global population, LAC more is decentralized) developed online platforms to ensure than triples that share when it comes to COVID-19-relat- education delivery. This was the most universal delivery ed global deaths - 28 percent (Figure ES.1). LAC has also channel. The second most used delivery method was TV, suffered from the largest hit in economic activity and pres- with roughly 85 percent of the systems offering this mo- ents the worst economic prospects in terms of recovery dality. Other learning channels used - including radio pro- than any other region. The economic decline had serious grams, social networks, short-message service (SMS) via repercussions for LAC, most notably through the sizeable cell phones, or printed material - were a bit less prevalent, impact on the labor market, especially from 2019 to 2020, but still widely used. Moreover, many governments scaled where unemployment rate for the region grew, on average, up support to teachers and encouraged the involvement of 53.6 percent. Last but not least, LAC has suffered from an parents and caregivers, as their participation and support inordinate intensity of an unprecedented phenomenon: for children’s learning during this lockdown period was the closing of educational institutions. LAC schools have essential. been fully or partially closed for a dismal 58 weeks, fea- turing as the third most affected region after South Asia Despite these commendable investments made to sup- and North America (Figure ES.2). In the region, an entire port learning continuity, the remote learning response generation of students – approximately 170 million – were faced several limitations, especially affecting the most deprived of in-person education for roughly 1 out of 2 ef- vulnerable countries and groups. The distance learning fective school days to date. response package faced several challenges, limiting its Executive summary Two Years After | Saving a Generation 19 Figure ES.3. Internet access across LAC, selected Figure ES.4. Attendance rates before and during the countries (%), latest available year pandemic, selected countries (%), February 2020 vs. May/July 2021 Haiti 4% 97% Chile 98% Guatemala 15% St Lucia Suriname 19% Costa Rica Honduras 20% Brazil Dominican 86% 21% LAC Republic 97% El Salvador 23% Dom. Rep. Peru 24% Haiti Colombia Total average 24% Mexico Guyana 24% Jamaica Paraguay 25% Honduras Mexico 34% Belize Colombia 38% Guatemala 64% Panama 42% Guyana 97% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% During pandemic Before pandemic Source: Own elaboration based on Barron-Rodriguez et al. (2021a). Note: Source: Own elaboration on the basis of WBG’s HFPS microdata library. Data for 20 LAC countries including the latest available data for each country. reach, participation, and quality. Five types of constraints the best scenario, remote learning was far from being a were particularly relevant in LAC: (i) the uneven state of perfect substitute to face-to-face education. connectivity in the region; (ii) the limited access to de- vices needed for distance learning, especially for vulner- The relatively limited available evidence so far on the able groups; (iii) the complexities in the implementation impact of the pandemic on schooling points to an im- of the specific response strategy, subject to the different portant attendance shock during remote learning for all strengths, limitations, and requirements of each distance children and youth, stronger for the most vulnerable. learning channel; (iv) the lack of teacher preparedness Before the COVID-19 outbreak, the out-of-school popula- for remote learning; and (v) institutional constraints. In tion (OOSP), both in absolute terms and as a share of the fact, recent data compiled for 12 countries in LAC show school age population, was consistently decreasing over that roughly only 1 in 4 households have access to Inter- time across education levels. Evidence collected from net (Figure ES.3) with noteworthy differences across coun- the High-Frequency Phone Survey (HFPS) showed that, tries. For example, while 40 percent of all households in as of mid-2021, when most of the schools in the region countries like Panama and Colombia report having Inter- were still fully or mostly closed, school attendance had de- net access, such indicator is below 15 percent in countries creased, on average, 12 percent from its pre-pandemic val- like Guatemala and Haiti. The quality of the connectivity ues (Figure ES.4), with important differences across coun- is also an important constraint for schooling and learning, tries. This decrease was sharper for lower-income groups. and vulnerable groups such as Afro-descendant students have significantly lower access to the Internet and a com- By the end of 2021, as schools were reopening, the puter at home, as compared to their non-Afro-descendant overall impact on schooling appeared limited in se- peers. Emerging regional evidence confirms that remote lected countries, with a very small increase in the 6-to- education delivery, especially with the multiple limita- 14-years-old OOSP population, but the risks of future tions mentioned above, has fallen short of ensuring en- dropouts for children and teenagers require continuous gagement in education activities at home, and quality of attention. By the end of 2021, in selected countries and learning. It has therefore become patent that, even under using other indicators, the overall impact on attendance/ Executive summary 20 Two Years After | Saving a Generation Figure ES.5. TERCE 2013, ERCE 2019 and simulated Figure ES.6. TERCE 2013, ERCE 2019 and simulated 2022 reading scores 2022 math scores 750 750 740 740 725 730 730 715 720 720 710 710 706 710 718 715 700 700 709 700 690 690 680 680 670 680 670 670 673 660 669 660 650 650 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022* 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022* Grade 3 Grade 6 Grade 3 Grade 6 Source: Own elaboration based on the UNESCO website and World Bank estimates based on Azevedo et al. (2022). Note: The country level mean scores between official UNESCO results and the baseline values for ERCE 2019 used in the simulation tool differ slightly because the simulation tool estimated baseline values for 2019 using grouped data. enrollments rates appeared limited. Reduced opportunity below MPL in reading and a dismal 83 percent were below costs of schooling, cash transfer policies, automatic pro- MPL in math. The arrival of the pandemic is expected to motion rates, among other factors, may have played a role have led to a crisis within a crisis. A strong negative impact in this outcome. Yet, data for four countries, represent- on learning is expected because of the limited reach, en- ing about 40 percent of the 6-to-14-years-old school age gagement in and quality of remote learning. population of the region, reveal an average small increase in the 6-to-14-year-old OOSP. Additionally, the relatively Simulations using observed data on the length of slow economic recovery, limited engagement and atten- school closures to date forecast large learning losses dance during school closures, mental distress and learning for the region, confirmed by early evidence. Updated losses may lead to large shocks in dropouts in 2022 and estimates using different assumptions on the extent of beyond, which could reverse the overall trends in OOSP, partial reopening and the effectiveness of remote learning for both children and teenagers and, especially, the most show a loss of between 1 and 1.8 learning-adjusted years vulnerable. of schooling (LAYS), with a loss of 1.5 LAYS under an inter- mediate scenario. These learning losses translate to a sig- While the ultimate effects on enrollment and drop-out nificant decrease in earnings and productivity equivalent may not be yet clear at this point, the pandemic is ex- to a decrease of about 12 percent in lifetime earnings for pected to have had a large impact on learning outcomes a student at school today under an intermediate scenario. and skills development, eroding the already weak foun- Under an intermediate scenario, mean ERCE scores are ex- dations of LAC’s education systems. Prior to the irrup- pected to decrease by around 6.3 percent (or 45 points) tion of COVID-19, the region was already characterized by in both grades and subjects. The simulation results for a deep learning crisis. According to results from the “best 3rd and 6th grade imply that every country would have pre-pandemic” measure of student achievement in the worse mean scores than in 2013. Put differently the mean region – ERCE 2019 results – the average student in LAC scores in reading and math would fall to levels of more suffered from a profound lack of core foundational skills, than 10 years ago (Figures ES.5 and ES.6), in a context namely literacy and numeracy. By mid-primary education where improvements were already very slow. In weight- (3rd grade), roughly half of the students in the region were ed terms, the proportion of 3rd and 6th graders not able below minimum proficiency levels (MPL) in both math and to adequately understand and interpret a text of moder- language; by the end of primary education, the situation ate length is expected to have increased, on average, from worsened significantly: 69 percent of 6th-graders were 37 to 50 percent and from 62 to 82 percent, respectively. Executive summary Two Years After | Saving a Generation 21 Figure ES.7. The road to recovery and acceleration in LAC RECOVERY FROM LEARNING RETURN TO SCHOOLING LOSSES AND ACCELERATION RECOVERING Safely Re-Enrolling Preventing Consolidating Assessing Implementing reopening dropouts the curriculum student’s learning recovery schools learning level programs • Addressing psychosocial needs • Addressing the digital gap And increases in learning poverty are the steepest in the unequal distribution of responsibilities in the household. world. According to the very latest World Bank estimates, In combination with the impact on schooling, these learn- the proportion of “learning poor” may have increased from ing losses lay the ground for a major generational crisis, if 52 percent in 2019 to 79 percent as a result of the pan- the right policies are not implemented, with a dual focus demic. Recent evidence collected on São Paulo, and more on schooling and learning recovery starting from a very recently on the Mexican State of Guanajuato, seems to fur- young age. This is even more true for the most vulnerable ther buttress the simulated estimates. While scores and students, who are at an especially great risk of dropping score scales are not comparable, percentage decreases out in the near future without urgent learning support, in primary education (ranging from about 8 percent to 19 with huge potential implications for inequality. percent) are comparable or even sharper; and grade 5 re- sults for São Paulo are equivalent to scores from 10 years The pandemic has also affected psychosocial health ago or more. While evidence is very scarce, transferable and well-being of teachers and students, while show- skills, which include skills like problem-solving, resilience, ing the importance of digital skills. Recent evidence on and communication skills and are shown to influence the the post-pandemic points to a deterioration of well-being ability to accumulate skills in the future, including foun- of children and teenagers across education levels, hinder- dational ones, have also have been negatively impacted. ing their readiness to learn and pointing to a mental health crisis within the crisis. At the same time, the pandemic also On average, early grades, younger children, and chil- revealed large gaps in digital skills, while providing some dren from lower socio-economic status have been dis- opportunities for enhancing those skills. More than ever proportionally affected by learning losses according before, education systems and schools were pushed to to early evidence, setting the stage for a generational adopt tools that implied a certain degree of digital ability. crisis and increased inequality. Results from São Paulo This caused great strain on teachers, students, and par- show much starker declines for grade 5 than for grades 9 ents, while also enhancing exposure to these skills and and 12, and data from Mexico also show steeper drops in opportunities for strengthening tools and strategies to de- primary than secondary education. The still very limited velop these critical competences. evidence on pre-primary education also points to import- ant losses. Data from Mexico also show steeper declines While recovery has started, the gravity of the crisis re- for lower-income students, with losses of 32 percent in quires comprehensive and sustained action, and this re- math scores for lower-income students vis-à-vis 25 per- port attempts to make a significant contribution to this cent for higher-income students. And other evidence in crusade by providing clear directions to education poli- the region and outside points to sharper decreases for the cymakers in LAC, building on the response so far. The most vulnerable groups. Female students also appear to document builds on a previous similar effort one year ago, have been more impacted by learning losses, suggestive of this time gathering three international organizations like the Executive summary 22 Two Years After | Saving a Generation Figure ES.8. The four commitments of the education recovery in LAC Learning and well-being: Schooling: Recover and enhance Teachers: Advocacy and nancing: Leave no one behind foundational skills and Value and support Place education recovery and prevent droputs well-being teachers at the top of the public agenda • Reopen all schools in a • Consolidate curricula with • Address human resource • Fund schools’ safety sustainable way. focus on foundational (and gaps and strengthen protocols, digital upgrades, • Support inclusive information transferrable) skills. teacher professional and all programs and campaigns on school • Assess learning levels, with development. measures at scale. re-enrollment, deploy cash summative and formative • Strengthen teachers' • Mobilize multiple transfer programs and ensure assessments. pedagogical and digital stakeholders in the delivery learning materials, informa- • Scale up initiatives and skills. and nancing e ort. tion, and services are programs to recover from • Support health and • Look for e ciency gains in accessible to all, especially learning losses, with focus well-being of teachers. use and allocation of those most vulnerable. on teaching at the level of resources, through better • Implement early warning the student and building on data, technological systems to identify and pre-existing and new innovations and institution- monitor students at risk of strategies and programs. al reforms for service dropping out. • Address psychosocial delivery improvements. • Address psychosocial health health and well-being of and well-being of students. students. • Address digital divides, with a • Address digital divides, focus on connectivity, digital with a focus on connectivi- skills and institutional ty, digital skills and capacity. institutional capacity. World Bank Group (WBG), the United Nations Children’s Fund continuing to address the digital divides that have been (UNICEF), and the United Nations Educational, Scientific and hampering the response. Embedded in all these priorities Cultural Organization (UNESCO) which, partnering with the is the support to teachers and principals, and ensuring ad- Inter-American Dialogue (IAD), join forces here to elevate the equate and efficiently used resources, to make their imple- pitch (and underscore the sense of urgency) for the sizeable mentation possible. schooling and learning recovery agenda ahead. In fact, if ur- gent and sustained action is not taken, an entire generation Returning to schooling and recovering from learning may suffer profound and long-lasting consequences on their losses entails the implementation of four commit- human capital accumulation. ments, aligned with this agenda (Figure ES.8) and the priorities put forward by the WBG, IAD, UNESCO and The main policy recommendation is clear: recovery UNICEF. These commitments include: (i) a commitment must focus on two essential strategies, returning to to schooling – so as to ensure that no student is left be- schooling and recovering from learning losses (Figure hind and/or drops out; (ii) a commitment to learning and ES.7). The “return to schooling” aims at ensuring, as a first well-being– to prioritize foundational skills and foster ap- priority, that all schools re-open and that all children who propriate levels of human capital formation along the way; were attending education prior to the pandemic can get (iii) a commitment to teachers – to make sure teachers are back to, and stay in, school. Two key reasons substanti- valued and supported throughout; and (iv) a commitment ate the key rationale for this strategy. First and foremost, to advocacy and financing – insofar as the education recov- COVID-19 poses limited health risks for younger popula- ery agenda is a responsibility of all and needs sufficient, tions. Second, foundational skills are not adequately de- and wisely used, resources for its implementation. The re- veloped at home. Once students are in school, they also port provides an actionable roadmap to ascertain which need to be learning and recover from the massive learning are solid measures, programs, interventions and strate- losses, therefore actions also need to be taken to offset gies to achieve the three first commitments, while flagging those losses and accelerate (“recovery from learning loss- risks and opportunities to achieve the fourth commitment. es and acceleration”). To be successful, this dual schooling and learning agenda will require addressing the psycho- The first step to support a return to schooling is to get social challenges triggered by the pandemic, as well as back to in-person education by finalizing the school Executive summary Two Years After | Saving a Generation 23 reopening process and make it sustainable. To ensure and characteristics of learning losses; and (iii) building on a safe return to in-person education, it is essential to ap- the curricular decisions and on the evaluation of learning ply context-appropriate health and hygiene protocols and losses, interventions that can help provide the needed (and to prioritize teachers in the vaccination process. School presumably accelerated) learning recovery. Underlying in- reopening strategies need to be coherent, flexible, and stitutional and service delivery constraints will also need clearly owned by, and communicated to, all stakeholders. to be addressed, and countries should as much as possible In the process of prioritizing school reopening, opening prioritize existing practices and programs, improving and early childhood education and primary education institu- scaling them up building on emerging lessons. tions should be a top priority, for all the above. Countries should keep prioritizing foundational skills. Open schools are not enough to ensure schooling: Building on efforts made so far during the pandemic, they countries need to make sure they have in place a set should focus on foundational, and transferable, skills in of smart policies to spur enrollment and attendance. their curricula in the short and longer term, and proficien- While enrollment and attendance data are still limited, cy in these skills should be measured. The pandemic may all efforts must be made to ensure re-enrollment and pre- be an opportunity to reiterate the importance of transfer- vent dropouts in 2022 and 2023, and prevent current or able skills and how to enhance them across the life-long future disconnects with schooling, building on what has learning cycle, while refocusing teaching and learning on worked so far, and with special emphasis on the most foundational skills. vulnerable children and youth. Three main interrelated reasons will make these efforts especially critical: (i) op- Learning assessments should be urgently re-priori- portunity costs of schooling may increase again, follow- tized, with a focus on formative assessments. Comple- ing changes in the labor market; (ii) financial constraints mentary assessment efforts need to be urgently stepped may remain binding, or even increase as focus goes back up everywhere to diagnose the real learning levels of stu- to face-to-face education, because of the slow recovery; dents, including recent learning losses, with a strong focus and (iii) learning gaps accumulated during school clo- on formative assessments. In the last decade, countries sures will become fully apparent. In this context, the fo- like Brazil, Chile, Mexico, and Uruguay introduced assess- cus should be on smart policies to spur re-enrollment and ments that are made available for the schools and teach- attendance. Building on good practices reviewed in this ers to administer, score, analyze, and use to improve ped- report, interventions should build on different incentives, agogical practices, and some other countries have taken with emphasis on keeping and enhancing the demand for initial steps along this line. Formative assessments have schooling. These interventions could include information the potential to be integrated with pedagogical approach- campaigns for re-enrollment; scaling up cash-transfers, es, measure transferable skills and to be replicated and softening their conditionalities; ensuring accessible mate- scaled up at a relatively low cost. At the same time, in- rials and services for all; and targeting additional support ternational, regional and/or national standardized assess- for specific vulnerable groups, while ramping up policies ments should continue to be fostered to generate compa- to recover from learning losses. Examples of effective cash rable data for benchmarking learning losses, and inform transfer policies can be found in many countries in the re- educational policies, in the context of a strengthened eval- gion, including Brazil, Mexico, and Colombia. These poli- uation culture where assessments work together to ulti- cies should be complemented by early warning systems to mately improve learning. identify the students most at risk of dropping out. Learning recovery programs should be scaled up. Learn- First and foremost, the return-to-schooling strategy ing losses are expected to be widespread, and learning must be accompanied by an aggressive strategy to re- heterogeneity to have increased, requiring a scale-up of cover and enhance foundational skills. In view of the reforms, initiatives, and programs to recover. This can massive learning losses, and expected impacts on drop- build on evaluated pre-existing programs (Figure ES.9), outs, this is an even more burning priority than school- and new programs introduced as a response to the pan- ing itself at this stage. This strategy must include three demic, to the extent that they start being monitored and critical priorities: (i) curricular consolidation with a focus evaluated. All programs should focus on foundational lit- on foundational, and transferable, skills, stating what stu- eracy and numeracy skills. Good monitoring and evalua- dents should learn; (ii) assessment of students’ current tion systems and strengthening institutional capacities to learning levels to properly diagnose the breadth, depth, implement at scale will be critical. Executive summary 24 Two Years After | Saving a Generation Countries willing to implement learning recovery pro- Figure ES.9. Examples of learning remediation and grams at scale will need to pay attention to critical recovery programs in LAC questions and features which are conducive to scalabil- TaRL ity. While information on unit costs is missing for most Teaching at the Right Level Mexico JEE types of programs, based on critical features to adapt and Extended Teaching Time scale up and evidence of results, programs like “Teaching Dominican Republic at the right level” may be especially promising to both recover and accelerate, but other programs may be less demanding in terms of training requirements. Targeted in- Early Learners Leveling. struction has strong potential to also accelerate learning in Caribbean the long run. Supporting a prompt recovery in early child- hood education will help tackle lack of readiness to learn. Aula Global Tutoring. Colombia Recovering and enhancing schooling and learning will also require a strong focus on psychosocial health and well-being. Countries should put in place diagnostic as- sessments to provide timely data on the effects of the pan- demic on psychosocial health and well-being and design Acelera Brasil strategies to address them. Some countries such as Chile Acceleration. PAM Brazil Computer-Assisted and Ecuador have strived to promote balanced recovery Learning approaches which include aspects related to well-being. Uruguay Source: Mancebo and Vaillant (2022). Recovering and enhancing learning will also require a strong focus on digital divides. At the same time, efforts to connect families and schools should continue, with fo- cus on both the access and quality of the Internet access, becoming an educational tragedy with large economic and which are still very uneven in the region. This needs to social implications. The region needs to protect education be combined with proper equipment and devices and the budgets first. The schooling and learning recovery package enhancement of digital skills of students and teachers and has a “price tag”. Improving school safety has a cost and so entail the needed institutional strengthening. Periodic as- implementing at scale programs to support attendance and sessments of digital skills should be supported. recover from learning losses. It is estimated that $47 billion is required to connect the region to the Internet for educa- Teachers need to be supported throughout the recov- tional purposes by 2030. Notwithstanding the primary role ery effort. Scaling up diagnostic assessments and learn- of ministries of education and public funding, the response ing recovery programs will require sufficient teachers and can also build on the many partnerships and alliances be- strengthening teacher professional development. Teach- tween ministries, the public and private sector, and govern- ers, tutors, and other implementers need to be trained and ments and civil society developed during the crisis. Further supported on how to target instruction, by teaching at the strengthening and developing these partnerships for educa- right level of students. Teachers’ digital skills also need tion is an opportunity provided by the crisis. Finally, there is to be strengthened to make the best use of technological also potential for efficiency improvements, to complement solutions, including through digital competencies frame- additional funding efforts. Countries can build on data and works. Finally, as first responders, teachers also need to be technology to improve service delivery through strength- trained to provide psychosocial support to students and ened information and monitoring and evaluation systems, their own health and well-being need to be supported. platforms to enhance targeted learning or, yet innovative and sustainable uses of remote and hybrid learning. While Finally, countries also have an important advocacy and the agenda of needed institutional reforms is much larger, funding agenda ahead of them with opportunities for the pandemic crisis and response has also shown how im- partnerships and efficiency improvements. A broad- portant it is to support principals and align incentives and based commitment and response are essential to avoid that capacities across stakeholders to implement successful re- the exogenous shock suffered by the 170 million students sponses. Countries could also capitalize on these efforts to in LAC that went through a real educational turmoil end up use resources more efficiently. Two Years After | Saving a Generation 26 Two Years After | Saving a Generation PHOTO BY: © MICHAL KNITL / SHUTTERSTOCK.COM Introduction Introduction Two Years After | Saving a Generation 27 I t has been over two years since the COVID-19 pan- demic first hit the Latin America and the Caribbean (LAC) region. What initially began as a health crisis of un- a sense of urgency, the report aims at providing an update on the impact of the pandemic on the education sector, extract- ing lessons on the response so far and building on those for known origin and consequences eventually shook the very an agenda for action. foundations of a region with structural weaknesses and a high level of inequality in both opportunities and income. The report builds on a previous similar effort one year The health crisis led to a spike in contagions and deaths, ago but joining efforts, this time around, to amplify mes- disproportionately affecting the region. LAC’s contribution sages for impact. This study is born as a series of concerted to worldwide deaths due to COVID-19 was far larger than actions taken by a group of institutions devoted to investing its share of worldwide population. The health crisis led to in human capital to build stronger and resilient societies. an immediate emergency situation for the economy with The World Bank Group (WBG), the United Nations Chil- the sudden and massive shutdown to normal life in order dren’s Fund (UNICEF) and the United Nations Educational, to prevent the spread of the virus. Transport, businesses, Scientific and Cultural Organization (UNESCO) join forces tourism: all came to a completely and sudden stop. And here to deliver a document that could help provide clear education, the most critical component for building hu- recommendations for the sizeable schooling and learning man capital and skills for life, was not an exception. recovery agenda ahead. In doing so, the three institutions partner with the Inter-American Dialogue (IAD) for advoca- The impact of the pandemic to the education sector was, cy through a series of webinars that help disseminate these however, of a different nature: it led to a silent crisis. Ed- efforts through their superb network of educational policy- ucation did not show people dying intubated in over-crowd- makers, practitioners, and academicians. ed and poorly resourced health institutions. Education did not mimic the massive loss of jobs and income for many The report has a simple but powerful structure designed people, especially the most vulnerable populations. But it to use the available evidence to date to substantiate engendered a tragedy whose real costs to the generation of the roadmap offered as a critical tool moving forward. students attending schools back in March 2020 will only be Chapter 1 is diagnostic in nature. The chapter attempts to fully felt in probably decades. The voices of the 114 million take stock of the 24+ months since the virus outbreak in students that were, only one year into the pandemic, still un- the region by dissecting the core elements of the health, able to receive in-person education, were never duly heard. economic, and educational shocks generated by the pan- And the size of the damage has not been fully assessed yet. A demic. When zeroing in on the characteristics of the dis- silent crisis, a generation at risk. ruption stemming from COVID-19, the chapter focuses on two core aspects: (i) the specific dimensions why the LAC This report is an attempt to raise awareness about “the region could be considered an “outlier” region; and (ii) great equalizer that vanished1” and to continue ensuring the particular dynamics of the remote learning response. a sense of urgency for addressing the crisis, while provid- Chapters 2 and 3 focus sequentially on two core outcomes ing clear directions to education policymakers in LAC on of the educational service delivery: schooling and learn- how to move forward building on the response so far2. The ing, while also looking at some other outcomes, with focus education process, usually subsumed within the two critical on skills development. Both chapters adopt a similar ap- outcomes of schooling and learning, is in fact a fundamental proach first presenting an overview of the pre-pandemic vehicle for skills development which simply cannot be put “on situation, to then portraying a data-based assessment of hold”. School helps build foundational skills, like critical liter- the impact of the pandemic to date. Chapters 4 and 5 mim- acy and numeracy abilities for functional mind growth. But it ic the treatment of their predecessor chapters but focusing is also the house where to build critical competences for life, on the schooling and learning recovery (and acceleration) from transferable skills to the new set of 21st century skills, agenda. Again here, the treatment attempts to distill the like digital and green skills. Most importantly, along the skills critical elements that should be prioritized in the imple- formation process, the equalization of opportunities that mentation of this strategy. Chapter 6 concludes with a takes place is paramount. While the recovery has started, it is summary of the key policy recommendations and critical yet too timid, and the task ahead is daunting. Underpinned by commitments moving forward. 1 See the blog by Jaime Saavedra Chanduvi on this at https://blogs.worldbank.org/education/schools-great-equalizer-vanished. 2 Even though all levels of education were deeply affected by the pandemic, this report focuses on compulsory education (pre-primary, primary and secondary). Nonetheless, many of the challenges and policies discussed in this document are also relevant to post-secondary/tertiary education. 28 Two Years After | Saving a Generation PHOTO BY: © 2022 FERNANDA_REYES/SHUTTERSTOCK Chapter 1 The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response Two Years After | Saving a Generation 29 M ore than two years have elapsed since the initial outbreak of, arguably, one of the most disruptive global threats to humankind in the world’s history: the Figure 1.1. Total cumulative number of deaths per million people, by region, 2020 - 2022 3,000 COVID-19 pandemic. The SARS-COV-23 variant of the coronavirus, first identified at the end of 2019 in the city 2,515 of Wuhan (China), was declared officially a worldwide pan- 2,500 demic on March 11, 2020. Ever since then, countries all over the planet were, for once in the history of the world, 2,000 united around the same unique priority: combating the vi- rus. To address this crisis, countries all around the world 1,500 and in LAC had to resort to multiple strategies to contain the pandemic. In education, this translated into a com- 1,000 plete shutdown of educational institutions, with countries having to change, from one day to the other, the delivery modality of education from in-person to remote and de- 500 sign a strategy to do so. This chapter sets the stage for the report by reviewing the health, economic and educa- 0 NA LAC ECA MENA SA EAP SSA tion consequences of the pandemic in LAC and the core distance learning response package that countries of the Source: Own elaboration on the basis of information from WHO COVID-19 region had to put together to maintain learning continuity Dashboard. Updated as of March 2, 2022. during school closures. countries – Brazil, Mexico, Argentina, and Colombia - were amongst the top 20 countries with the largest number of 1.1 A health and economic crisis people infected in the world.4 The region was also in third like no other…disproportionately place when adjusting the contagions indicator by popula- affecting LAC tion size (i.e. cases per million people). The sanitary pic- ture is, however, far worse when we analyze COVID-19-in- Countries all around the world have had to set up all duced deaths. LAC is the second region with the largest sorts of strategies to contain the spread of the disease. cumulative number of deaths to date (Figure 1.1) with a In the absence of a vaccine, the most widespread modality small difference to the first one (North America/NA), locat- to halt transmission rates was the application of a gener- ed in the same continent. In fact, while being home to only alized strict lockdown. Heavy restrictions limiting move- eight percent of the global population, LAC more than tri- ment, economic activities, and social services were mas- ples that share when it comes to COVID-19-related global sively imposed in all corners of the globe. Although these deaths - 28 percent. The LAC region presents the second measures had varying degrees of effectiveness to curtail highest ratio between the share of total deaths and the transmission rates, they were not effective in ending the share of total population, a dismal 3.3 value (Figure 1.2). pandemic altogether. In fact, even in the light of substan- Taken in conjunction with the NA region - comprising the tial progress in vaccination to date, the pandemic contin- United States and Canada - the Americas is the continent ues to wreak havoc with waves of contagions and deaths. that has been, by far, the most lethally affected to date. As of today, one LAC country, Peru, heads the world ranking The pandemic hit almost everywhere, but some regions with the largest number of deaths per million people. And of the world, notably the LAC region, were dispropor- even when considering excess mortality due to COVID-19, tionately affected. As of the beginning of March 2022, new evidence points out to Bolivia, another South Amer- LAC was the region with the third largest total number of ican country, as being the country with the largest excess COVID-19 registered cases. The four most populous LAC mortality rate, and the Andean countries (Bolivia, Peru, 3 The SARS-CoV-2 acronym stands for severe acute respiratory syndrome coronavirus 2. This virus causes respiratory illness, is highly transmissible, and may eventually lead to death. Fatality rates are disproportionately higher for: (i) the elderly (ages 70+); and (ii) individuals with comorbidities. 4 For daily updates, see the COVID-19 Dashboard from the World Health Organization (WHO), at https://COVID19.who.int/. Latest update for this report is as of March 2, 2022. The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response 30 Two Years After | Saving a Generation Figure 1.2. Contribution to global totals for Figure 1.3. Share of population fully and partially population and number of COVID-19 related deaths, vaccinated, by region (%), 2020-2022 by region (%), 2020-2022 East Asia 79% 4% Sub-Saharan Africa and the Paci c North America 67% 11% East Asia and the Paci c Latin America and the Caribbean 68% 9% South Asia South Asia 59% 10% Middle East and North Africa Europe and Central Asia 63% 3% Europe and Central Asia Middle East Latin America 42% 7% 28% and North Africa and the Caribbean 8% Sub-Saharan Africa 13% 5% North America 0% 10% 20% 30% 40% 50% 60% 70% 80%90% 0% 5% 10% 15% 20% 25% 30% 35% Share Fully Vaccinated Share Partially vaccinated Contribution to Deaths Contribution to Population Source: Own elaboration on the basis of information from WHO COVID-19 Source: Own elaboration on the basis of Our World in Data, April 2022. Dashboard. Updated as of March 2, 2022. Ecuador) as having been disproportionately affected by more successful in acquiring vaccines and vaccinating its excess mortality due to COVID-195. population when compared to countries in the Caribbean.8 Whilst the death toll in LAC has been devastating, some The health crisis was just the tip of a giant iceberg that countries in the region have regained control over the permeated all areas of people’s lives, most notably immediate health crisis through increased access to their livelihoods, and here again, LAC was the region vaccines and significant vaccination rates. As of the end most severely impacted. In fact, when comparing across of April 2022, 68 percent of the population in the region world regions, we observe that LAC faced a double curse: has been fully vaccinated6, and an extra 9 percent partial- not only has the region experienced the most severe eco- ly vaccinated against COVID-19 (Figure 1.3). Compared to nomic recession to date due to the pandemic, but pros- other regions, vaccination rates were low at the start of pects for recovery in the coming years are projected to be the pandemic, but greatly increased during the last year the lowest. In fact, as Figure 1.4 shows, taking 2018 as the to have a larger share of the population vaccinated. Most base year for measuring the evolution of the growth rate importantly, however, average regional vaccination figures of the Gross Domestic Product (GDP), LAC is the region conceal a large variation among countries. While countries that experienced the largest decline in 2020 and where like Chile and Cuba - with 91 and 88 percent of people ful- forecasted rebound through 2023 will only be at 5 percent ly vaccinated, respectively - are atop of the regional rank- of GDP annual growth, 5 times lower than, for example, the ing, countries like Haiti - with one percent of its population East Asia and the Pacific (EAP) region. The world economy vaccinated – and Jamaica – 23 percent - are very far be- is now in a worse position than what was previously ex- hind.7 In general, vaccination rates in LAC show that larg- pected. In addition to a global economy that has not fully er countries in the South America sub-region have been recovered from the pandemic, Russia’s invasion of Ukraine 5 The excess mortality due to COVID-19 is a concept that has been used in the literature to document the total number of deaths, directly or indirectly, attribut- able to the pandemic. For the latest systematic review and data compilation from around the world to date, see COVID-19 Excess Mortality Collaborators (2022). 6 Full vaccination, in this report, refers to the application of the two doses of the vaccine. Partial vaccination refers to one dose. 7 See WBG COVID-19 vaccine deployment tracker at https://COVID19vaccinedeploymenttracker.worldbank.org/tracker. Another very useful complementary source is “Our world in data” at https://ourworldindata.org/. 8 On August 11, 2021, the Pan American Health Organization (PAHO) issued a statement - https://www.paho.org/en/news/11-8-2021-paho-director-appeals-ca- ribbean-people-get-vaccinated-observe-protective-measures - addressing the concerning vaccine hesitation in the region, with specific reference to the Caribbean, causing new spikes and outbreaks of the virus. The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response Two Years After | Saving a Generation 31 Figure 1.4. Index of real GDP growth rate, by region, Figure 1.5. Increase in the unemployment rate during 2018-2023 (est.) the pandemic, by region (%), 2019-2020 and 2019-2021 1.30 60 53.6 1.25 50.3 50 1.20 39.9 1.15 40 37.9 33.9 32.9 1.10 30 1.05 23.2 21.2 19.6 1.00 20 14.3 0.95 11.2 10 9.1 0.94 0.90 2018 2019 2020 2021e 2022f 2023f 0 EAP ECA LAC MENA SA SSA LAC NA EAP SA MENA SSA ECA Source: Own elaboration on the basis of WBG’s Global Economic Prospects, Source: Own elaboration on the basis of IMF’s October (2021) World Econo- January (2022) (https://www.worldbank.org/en/publication/global-econo- mic Outlook Database (https://www.imf.org/en/Publications/WEO/weo-data- mic-prospects). base/2021/October). has worsened these economic forecasts which are now ac- 1.2 The COVID-19 pandemic and its companied by high inflation rates. impact on LAC’s education systems to date The economic fallout had serious repercussions for LAC, most notably through the sizeable impact on the labor The COVID-19 pandemic triggered not only a worldwide market. Between 2019 and 2020, LAC was the region with health and economic crisis, but a profound shock to the the largest increase in unemployment rates with a dismal education sector. Ever since the pandemic was declared 53.6 percent rise in just one year (Figure 1.5), reaching an in mid-March 2020, there was one aspect where education 11.5 percent average rate of unemployment for the region.9 systems around the world overlapped fully: their uniform re- Only when comparing the latest estimates for the two-year action. Almost all11 countries around the world mandated the period comprising 2019-2021, does LAC not come on top. complete shutdown of educational institutions (henceforth, Still, even this two-year average shows an increase in the “school closures”12) and the sudden move of the delivery of unemployment rate of 32.9 percent, just one percentage education to distance or remote learning13. Interestingly, the point shy of that for the South Asia (SA) region. In a nut- seemingly equal response to a fairly similar timing of the shell, LAC is still an “outlier of the pandemic”, a region pandemic (on or around March 2020) generated an unequal where negative effects got exacerbated, leaving deep scars effect on regions across the world, depending on: (i) which on economies and societies in what was already the second hemisphere they were located in; and (ii) how their school most unequal region in the world prior to the pandemic.10 calendars were organized. Since LAC is the only region whose geographical extension (from Argentina and Chile in the South to Mexico and The Bahamas in the North) spreads across almost equally between the Northern and South- ern hemispheres, countries’ school calendars vary widely 9 See IMF (2021). 10 According to the latest update from the World Inequality Database, the most unequal region in the world is the Middle East and North Africa (MENA) region, with LAC coming in second place. See https://wid.world/news-article/2020-regional-updates/. 11 A handful of countries did not follow this pattern. In LAC, Nicaragua for example decided not to close schools at all. 12 The pandemic led to the generalized shutdown of all sorts of educational institutions (including early childhood, primary, secondary, and post-secondary ed- ucation institutions). For simplicity, this book refers to this phenomenon with the expression school closures. 13 Remote learning refers to the teaching and learning process that takes place at a distance, away from the physical school environment. It is a shift from the traditional face-to-face learning. In this report, the expression “distance learning” is used interchangeably with the more established “remote learning”. The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response 32 Two Years After | Saving a Generation Figure 1.6. School calendars in LAC and timing of the COVID-19 pandemic: Understanding the implications Number Countries March 2020 Jan: El Salvador, Guatemala, Nicaragua Feb: Bolivia, Colombia, Costa Rica, Pandemic hit at  Whole Honduras (2) (public schools), Paraguay the start of the  14 year 2020 school year Mar: Argentina, Brazil, Uruguay, Panama, Chile, Peru May: Ecuador (1) (Coastal Region) Jul: Curaçao, Sint Maarten (Dutch part), Saint Martin (French part) January Aug: Aruba, Colombia (2) (private schools), Dominican Republic, Guyana, Honduras (2) February Pandemic hit at  (private schools), Puerto Rico, Suriname the end of the  March Split year 34 Sept: Anguilla, Antigua and Barbuda, 2019-2020 April Bahamas, Barbados, Belize, Bermuda, school year British Virgin Islands, Cayman Islands, Cuba, May Dominica, Ecuador (1) Sierra region,  Grenada, Jamaica, Mexico, Montserrat, Saint June Kitts and Nevis, Saint Lucia, July Saint Vincent and the Grenadines, Trinidad and Tobago, Turks and Caicos Islands, August US Virgin Islands,  Venezuela September October (1) Ecuador has different school calendars depending on the region - May-Feb for the Coastal region; Source: Own elaboration on the basis of UNESCO Institute Sept-June for the Sierra region. (2) Honduras and Colombia have different school calendars depen- of Statistics (http://data.uis.unesco.org/#) and https://elor- ding on the type of schools – Feb-Nov for the public sectors; Aug-May for the private sector. denmundial.com/mapas-y-graficos/inicio-del-ano-escolar- en-el-mundo/. Discrepancies between the table and the map may exist due to the averaging process for each data source. (Figure 1.6). Hence, while most countries in the Southern pandemic show LAC as the region with the largest length of Cone – from Chile in the West to Brazil in the East and many full school closures, alongside the SA region. On average, Central American countries – had just started their school LAC students had their schools fully closed for a dismal year 2020, many others in the Caribbean and some in Cen- 29 school weeks, roughly 7 full months. When considering tral America were starting their last quarter of school year partial school closures16, LAC has had a combined average 2019/2020. Overall, whereas all 45 countries in LAC had to length of 58 weeks, well exceeding a full year of learning stop and move to distance learning, students in one-third of and in-person interaction between students and teachers these countries (14) were hit at the beginning of their whole- (Figure 1.7). When disaggregating school closure statistics year school year, while those in the other two-thirds (3414) even further, we find that, on average, LAC schools were were hit at the end of their split-year school year15. closed (or mostly closed) for half of the effective number of school days in the last 2 years (47%). This appalling School closures occurred globally following the out- number increases to 64 percent when considering partial break of the pandemic but were especially long in school closures (Figure 1.8).17 LAC. Beyond the challenges with different academic cal- endars across the region, more than two years into the 14 Given that Ecuador and Honduras have dual school calendars, they are taken as two different school systems/countries for the purpose of the calculation of the number for split-year school calendars. 15 Difference in school calendars, given the timing of the pandemic, gave countries with split-year calendars a bit of an edge as they had the possibility of bear- ing only a partial-year impact of distance learning before entering the summer break, giving them a few more months to adjust with schools closed due to the aca- demic break before entering their new school year 2020/2021. 16 The “partially closed/open schools” category in the UNESCO data is defined as schools that are neither fully closed nor fully open or on academic break. 17 Reporting accurate data and information on the type and intensity of school closures has been challenging for many different reasons, including the unpre- dictability of the pandemic and the various government responses. Additional challenges lay in the definition of “partially open” school systems and the monitoring of the actual reopening process. There exist variations in the degree of the reopening beyond the “open”, “partially open” and “closed” systems with varied impact on schooling and learning that warrants a differentiated analysis. Further, many countries have rolled out the reopening of schools in different phases, geographic areas, and grades. Last but not least, the actual reopening of schools may have not necessarily mean that students attend. This report has undertaken a massive effort to provide a detailed picture for both the type and the intensity of school closures. The methodology is described in Annex 1.2. The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response Two Years After | Saving a Generation 33 Figure 1.7. Number of school weeks with schools fully During the time already elapsed since the start of the and partially closed, by region, March 2020 - March pandemic, progress in the school reopening process has 2022 not been linear and has been closely linked to the dynam- ics of the health crisis. At the beginning of the pandemic, North America 7 56 with a virus deemed lethal, unknown, and for which no vac- Europe and cine existed at the moment, most countries in LAC decided Central Asia 13 15 to close schools and offer distance learning. By the end of East Asia and March 2020, 98% of LAC countries that were not on aca- the Paci c 16 17 demic break had their schools completely closed18, affecting Sub-Saharan Africa 18 13 more than 95% of the total enrollment in compulsory edu- Middle East cation in LAC. During the second semester of 2020, schools and North Africa 24 21 progressively began to open. By December 2020, roughly Latin America half of the countries not on academic break (largely Carib- and the Caribbean 29 29 bean countries) still had schools fully or mostly closed19. At South Asia 35 31 the start of 2021, a second wave of COVID-19 cases led to a temporary setback in the school reopening process. By 0 10 20 30 40 50 60 70 end-March 2021, exactly one year since the virus outbreak, Avg. weeks of Avg. weeks of full closures partial closures the situation in LAC resembled very closely that of 6 months back. During the remainder of 2021, with progress in vacci- Source: Own elaboration on the basis UNESCO’s Global Monitoring of School Closures. Updated as of March. 31, 2022. nation, including boosters (third doses) already in several countries, the school reopening process showed progress once again. By March 2022, 87% of countries had schools Figure 1.8. School closure intensity (%) by country, March 2020 - March 2022 (share of school weeks, by intensity of closure) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Peru Honduras El Salvador Ecuador Belize Jamaica Panama Guatemala Venezuela Chile Guyana Mexico U.S. Virgin Islands Puerto Rico Bolivia Colombia Barbados Brazil Dominican Republic Saint Lucia Costa Rica Cuba Dominica Paraguay Argentina Montserrat Bahamas Turks and Caicos Islands Suriname Antigua and Barbuda Trinidad and Tobago Grenada Saint Vincent and the Grenadines Saint Martin Anguilla Saint Kitts and Nevis Haiti Sint Maarten Cayman Islands British Virgin Islands Bermuda Uruguay Aruba Curaçao Nicaragua Fully Closed Mostly Closed Partially Open/Closed Mostly Open Fully Open Source: Own calculations based on UNESCO and UNICEF’s LAC COVID-19 Education Response Updates from March 1, 2020 through March 31, 2022. 18 Of the 45 countries in LAC, schools were completely closed in 40 countries, on academic break in 4 countries and open in one country. 19 Eleven countries were completely closed, 6 countries mostly closed, 4 countries partially open/closed, 18 countries fully open and 6 countries on academic break. The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response 34 Two Years After | Saving a Generation Figure 1.9. Dynamics of school closures and reopening in LAC countries, by end of quarter (% of total school weeks by status), March 2020 – March 2022 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% March June Sept Dec March June Sept Dec Mar 2020 2021 2022 Fully Open Mostly Open Partially Closed/Open Mostly Closed Fully Closed Source: Own calculations based on UNESCO and UNICEF’s LAC COVID-19 Education Response Updates from March 1, 2020 through March 31, 2022. fully or mostly open, with the remaining countries being ones to suffer the shortest, with countries in South Amer- partially open/closed (Figure 1.9). ica being somewhere in between. In fact, 5 of the top 10 countries of the School Closure Index – Belize, El Salvador, Whilst the average length of school closures for the re- Guatemala, Honduras, and Panama– are located in Central gion is concerningly long, as compared to the rest of America (Figure 1.11), whereas only 2 of the 10 countries the world, the variation in length across countries in with the shortest school closures – Nicaragua and Uru- LAC is large. The information provided in the previous guay23 – are not located in the Caribbean. paragraph has been calculated using a country-wide aver- age for LAC, estimated from a total of 45 countries in the Data on the length and intensity of school closures region. However, such average masks significant differenc- across LAC indicate that countries and governments in es. In fact, 7 countries in LAC had their schools closed for the region approached school closures and reopening in more than three-fourths of the net school time in the last different ways. As was shown, the majority of small island two years20, including Mexico, the second most populous states in the Caribbean were the ones to close less and/ country in LAC, with roughly 35 million students of com- or reopen earlier, arguably due to their insular situation, pulsory education age. Conversely, 11 countries had their which gave them an edge in terms of border closing and schools closed for less than one-fourth of the school time virus spread. Still, some specific Caribbean islands like The to date (Figure 1.10). The School Closure Index21, which Bahamas, Jamaica, Puerto Rico, and the U.S. Virgin Islands measures the intensity of the school closure in LAC coun- were among the top 15 countries with longest spells of tries, shows that, on average, countries in Central Ameri- school closures. The massive and pervasive school closures ca (including Mexico)22 were the ones to suffer the longest have been heterogenous in nature, not only across coun- school closure spells, and countries in the Caribbean the tries but similarly within. Many countries, including Bolivia, 20 Net school time is equivalent to the actual days of each country’s individual academic calendar excluding academic break. 21 The index was created using 5 categories of school closure, as follows: (i) “fully closed” (100% of schools closed); (ii) “mostly closed” (67%+ of schools closed); (iii) “partially closed/open” (ranging from 33% and 67% of schools closed/open); (iv) “mostly open” (67%+ of schools open); and (v) “fully open” (100% of schools open). The index then is built around a maximum value of 400 and a minimum value of 0, and is calculated as per the following formula: School Closure Index = (% of effective school days that schools were “Fully Closed” * 4) + (% of effective school days that schools were “Mostly Closed” *3) + (% of effective school days that schools were “Partially Closed/Open” *2) + (Share of effective school days that schools were “Mostly Open” *1) + Share of effective school days that schools were “Fully Open” * 0). 22 With the exception of Nicaragua that did not close schools. 23 In the case of Nicaragua, although all schools never officially closed, student attendance was irregular for several months due to fear of COVID-19 infection. Schools in Uruguay were reopened early, between May and June of 2020, and they remained open, except for April 2021. During remote learning periods, students participated actively through the educational platforms of the CEIBAL Plan. The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response Two Years After | Saving a Generation 35 Figure 1.10. Summary of school closures for LAC, by country (share of school weeks with schools fully or mostly closed to date) School Weeks with Closed or Mostly Number of List of countries Closed Schools (% of total) countries > 75% 7 Belize, Ecuador, El Salvador, Honduras, Jamaica, Panama, Peru 50-74% 13 Barbados, Bolivia, Brazil, Chile, Colombia, Dominican Republic, Guatemala, Guyana, Mexico, Puerto Rico, Saint Lucia, Venezuela, and Virgin Islands (US) 25-49% 14 Antigua and Barbuda, Argentina, Bahamas, Costa Rica, Cuba, Dominica, Grenada, Montserrat, Paraguay, Saint Martin, Saint Vincent and the Grenadines, Suriname, Trinidad and Tobago, and Turks and Caicos Islands < 24% 11 Anguilla, Aruba, Bermuda, British Virgin Islands, Cayman Islands, Curacao, Haiti, Nicaragua, Saint Kitts and Nevis, Sint Maarten, Uruguay Source: Own calculations based on UNICEF, 2021. LAC COVID-19 Education Response Updates from March 1, 2020 through March 31, 2022. Figure 1.11. School closure index: Top 15 LAC and placed emphasis on inclusive practices and provision countries of adapted learning materials to relevant groups. Some governments also prioritized school reopening for different Bahamas 278 grades. Evidence from the first year of the pandemic indi- Saint Lucia 280 cates that LAC countries, on average, tended to prioritize Puerto Rico 286 return to school for students at the secondary education Bolivia 288 level. This pattern was slightly different from a few other Virgin Islands (US) 293 regions, such as EAP, Europe, and Central Asia (ECA), and Jamaica 297 the Middle East and North Africa Region (MENA) that rath- Belize 299 er tended to prioritize return for pre-primary and primary Guatemala 301 levels (Figure 1.12). Further analysis of and lessons on the Venezuela 304 school reopening strategy are provided in Chapter 4. Mexico 310 Ecuador 318 Peru 320 El Salvador 326 1.3 Overview of the remote learning Panama 327 response to the pandemic Honduras 350 100 150 200 250 300 350 400 1.3.1 The elements of a complex remote learning response Source: Own calculations based on UNICEF, 2022a. LAC COVID-19 Education Response Updates from March 1, 2020 through March 31, 2022. School Closure Index = [Fully Closed (% of effective school days) * 4 + Mostly Education systems across the LAC region responded Closed (% of effective school days) *3 + Partially open/closed (% of effective to the irruption of COVID-19 and school closures in a school days) *2 + Mostly Open (% of effective school days) *1 + Open (% of effective school days) * 0]. strong fashion and undertook notable efforts to contin- ue delivering education remotely during the pandem- ic24. These efforts comprised several aspects that char- Chile, Ecuador, Mexico, Panama, Paraguay, Peru, and Uru- acterized the “distance learning response package”, from guay, initiated the school reopening process in rural areas the education delivery channels chosen to reach students with lower transmission rates and COVID-19 prevalence. from afar to the enhancement of needed investments to Other countries like Colombia empowered sub-national lev- ensure delivery and inclusion to strategies used for sup- els to reopen. Some governments also prioritized return to porting the different actors of the education process (stu- school for specific groups, including for vulnerable students dents, teachers, principals, households) along the way. 24 Such innovations are well documented in WBG (2021a). The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response 36 Two Years After | Saving a Generation Figure 1.12. Average number of instruction days lost during 2020, by level of education and region 250 200 150 100 50 0 East Asia and Europe and Central Latin America and Middle East and South Asia (N=3) Sub-Saharan the Paci c (N=11) Asia (N=35) the Caribbean (N=18) North Africa (N=6) Africa (N=23) Pre-Primary Primary Lower Secondary Upper Secondary Source: UNESCO, UNICEF, WBG, and OECD (2021b). As a result, the “remote learning experience during the Figure 1.13. Channels implemented for learning pandemic” to date has not been a uniform process, but continuity in LAC during 2020 (% of countries) rather a complex enterprise that attempted to accom- modate different education delivery modalities (largely Home visit 15% remote, and, to the extent that schools were at least par- Other tially opened, hybrid25), and channels – with a vast array platforms 20% of delivery backgrounds and situations. Hence, distance learning during the pandemic followed a multi-modality Paper 60% multi-channel multi-strategy approach. Such configuration SMS/social determines that there are several angles that are important media 60% to bear in mind so as to adequately understand the impli- cations of the sudden switch in the modus operandi of the Radio 70% delivery process. Beyond the core response package, sev- eral countries also developed additional policies and took TV 85% measures to support student retention and learning, such Government as early warning systems and curriculum consolidation, but platforms 100% these complementary efforts will be further examined in 0% 20% 40% 60% 80% 100% other chapters, to keep the focus here on the core response. Own elaboration based on Barron-Rodriguez et al. (2021a). Note: Data for only 20 LAC countries including the latest available data for One of the most important elements of the distance each country. learning response package was the delivery channel. Governments across the region were forced to combine learning materials, and even direct home visits were learn- the use of online learning platforms – the default modali- ing channels for many of the students in LAC. In a recent ty for remote learning – with other core elements or tools WBG study on the remote learning responses during the that could ensure teaching continuity for as many students pandemic carried out in 20 LAC countries, all of them had as possible. Hence, the usage of television (TV), radio pro- developed a government online platform to ensure contin- grams, usage of text messages/social media, take-home uation of education delivery (Figure 1.13).26 The second 25 Hybrid education implies any combination of in-person and remote learning, opening up an array of implementation possibilities that can be defined through three key features: i) time (when); ii) space (where); and iii) interaction (how). When planning for hybrid education, these three elements are critical aspects to consider when determining what type of teaching should be delivered remotely and what type in-person. See Barron-Rodriguez et al. (2021b). 26 Barron-Rodriguez et al. (2021a). The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response Two Years After | Saving a Generation 37 Figure 1.14. Support provided to teachers nationwide during 2020, by type of support and region (%) 120 100 80 60 40 20 0 Special Instruction Professional, Guidelines Professional Teaching Provision of Other Training on distance psychosocial, for e ciency development content for ICT tools and support instruction and emotional in remote activities remote connectivity support teaching teaching East Asia and the Paci c (N=14) Europe and Central Asia (N=38) Latin America and the Caribbean (N=27) Middle East and North Africa (N=9) South Asia (N=4) Sub-Saharan Africa (N=25) Source: UNESCO, UNICEF, WBG, and OECD (2021b). most used delivery method, country-wise was TV, with to other regions, and 82 percent of countries encouraged at 17 of these countries (85%) offering this modality. Other least four interactions or more between teachers and parents learning channels used - including radio programs, social and/or students during school closures (Figure 1.15).29 In this networks, short-message service (SMS) via cell phones, or regard, the LAC region appears to have been the second most printed material - were less prevalent, but still widely used committed region in the active pursuit of teacher-parent in- in the majority of the countries. teractions, behind the MENA region. Countries also invested substantially in teachers There were a few LAC countries whose distance learn- during the pandemic, albeit with a short-term focus. ing strategies were relatively successful at enhancing This included a variety of interventions, from ensuring participation. One of such experiences was Chile’s, which that teaching methods were adapted to the new remote showed that it is possible to utilize multimodal learning and hybrid learning environment, to investing directly in solutions effectively (Box 1.1). Chile implemented more additional teacher training, with specific emphasis on dig- than 50 education-related actions to respond to the pan- ital skills. Investments in professional development, dis- demic to ensure continuation and recovery of school- tance teaching instruction and professional, psychosocial, ing and learning, including a multimodal approach using and emotional support for teachers, were made by a large adaptive learning solutions, an online platform, broadcast majority of LAC countries (Figure 1.14).27 media (TV and radio) and printed materials, accompanied with critical support for teachers. Another good practice To complement investment in teachers, many govern- in distance learning education was Uruguay’s. This South ments also encouraged the involvement of parents and American country had established an educational technol- caregivers, as their participation and support in children’s ogy plan (Plan Ceibal) before the pandemic hit. This plan learning was critical. In LAC, communication between aimed at providing a personal computer to every student households and schools was most encouraged28, as compared in primary and middle public schools, Internet access to all 27 UNESCO, UNICEF, WBG, and OECD (2021b). 28 Beyond online learning interaction. 29 UNESCO, UNICEF, WBG, and OECD (2021b). The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response 38 Two Years After | Saving a Generation Figure 1.15. Number of interactions encouraged between teachers and parents and/or students during school closures in 2020, by region (%) 100% 90% 25 26 22 29 33 80% 50 70% 10 15 60% 21 11 43 50% 30 11 100 40% 21 41 30% 33 18 50 20% 14 35 11 10% 18 14 11 7 0% East Asia Europe and Latin America Middle East North America South Asia Sub-Saharan and the Paci c Central Asia and the Caribbean and North Africa (N=1) (N=4) Africa (N=23) (N=14) (N=40) (N=27) (N=9) No interactions Between 1-3 Between 4-6 Between 7-8 Greater than 8 Source: UNESCO, UNICEF, the World Bank, and OECD (2021b). schools, and a comprehensive set of educational resourc- detail below, followed by initial evidence on the impact of es and pedagogical services and programs. During the remote schooling in the region. COVID-19 disruption, Plan Ceibal adapted and strength- ened their services to teachers, students, and families, Arguably one of the most important challenges faced launching Ceibal en Casa (Ceibal at home).30 It is important by the remote learning package was the region’s defi- to note that both countries effectively capitalized during cit in connectivity. Recent data compiled for 12 coun- the pandemic on their pre-existing investment and institu- tries in LAC show that roughly only 1 in 4 households tional capacities. have access to Internet (Figure 1.16).31 Even within this important handicap for the region, there are noteworthy 1.3.2 The limitations of the differences across countries. While about 40 percent of remote learning response all households in countries like Panama and Colombia report having Internet access, such indicator is below 15 Despite the notable efforts and investments made by percent in Guatemala and Haiti. Access to TV and radio most countries in the region to support learning conti- is, as expected, higher in most LAC countries though still nuity, the remote learning response faced several lim- far from universal. Available data, using the same source, itations. Among the most important factors that posed a indicate that 81 percent of households of these 12 partic- challenge to the distance learning response package and, ipating countries had access to a TV and 70 percent to more generally, that have conditioned its reach, participa- a radio. Of all the platforms for remote learning access, tion, and quality, we find five relevant ones: (i) the state of mobile phones represented the most widespread one, connectivity in the region (quantity and quality); (ii) the slightly above access to a TV, with 81.3 percent of house- access to devices; (iii) the implementation of the response holds in LAC having access to this service. Limited access strategy; (iv) teachers’ preparedness; and (v) institution- to Internet has constrained access to remote learning.32 al constraints. Each of these factors is explained in more Internet access is also an issue in schools limiting op- tions for hybrid learning as schools reopen. 30 Students and teachers were given access to virtual learning environments, math platforms, a national digital library, as well as training and support. Families were provided content and guidance on how to support pedagogical continuity as well as socioemotional support. In addition to the digital resources provided by Plan Ceibal, paper-based resources and television broadcasts were made available during the COVID-19 disruptions. See Meinck, S. et al. (2022). 31 Barron-Rodriguez et al. (2021a). 32 Providing education through online platforms where students have limited access to Internet, instead of more low-tech solutions, has been coined as the re- mote learning paradox. This expression describes a scenario where students most at risk are unable to access learning material/resources provided through govern- ment delivery channels, as current digital infrastructure is not adequate for this purpose, which is unfortunately a reality in many countries across LAC. See Bar- ron-Rodriguez et al. (2021a). The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response Two Years After | Saving a Generation 39 Box 1.1. The remote learning experience in Chile During the school/calendar year 2020 alone, Chilean schools remained completely closed for 27 school weeks due to the pandemic. To mitigate the impact of the pandemic on education outcomes, the Government of Chile implemented a comprehensive multimodal approach to remote learning. The response consolidated more than 50 actions, cover- ing 8 main areas: (1) provision of resources and supplies for connectivity/education continuity; (2) distance learning strategy and channels; (3) reinforced or additional pedagogical materials; (4) flexibility and contextualization; (5) in-service teacher training; (6) strengthening of student retention and drop-out mitigation strategies; (7) socio-emo- tional well-being; and (8) safe return to face-to-face classes. The scope of the Chilean response combined with a great level of flexibility, allowing each institution to choose a combination of solutions best adapted to their context, resulted in relatively high levels of engagement and atten- dance. For example: • Aprendo en Línea (I learn online), an educational platform made available from the first day of suspension of face- to-face classes has had more than 13 million visits, with an average of more than 300,000 weekly users. • It was complemented by radio (Aprendo FM), TV (Aprendo TV) programs, and the “Aprendo en Casa (I Learn at Home)” program guarantying the same educational material in printed format, especially to rural schools and those with limited connectivity. • An agreement was also made with Athena, the Mobile Telephone Association of Chile to ensure free downloads of texts and access to their learning material, and with google and Fundación Chile to facilitate access and technical support. Their digital library has more than 10,000 free books available. Four strengths of the actions implemented by Chile, as compared to benchmarked countries were: • The response package was quick and high-quality. • Areas within the package showed a good overall alignment and coordination, leading to a coherent and compre- hensive approach. • Centralized proposals were offered with a great level of flexibility, allowing each educational institution to choose among those best aligned with their individual context. • Many of the actions implemented have the potential to become long-term strategies. Sources: MINEDUC-Chile and WBG (2022); and Bellei et al. (2022). Constraints are stronger for the most vulnerable needs33. Indeed, while access to electricity in LAC is al- groups. And this obstacle disproportionately affected spe- most universal in most urban areas, only one in every 10 cific groups insofar as access to Internet and electricity, households in rural areas has access to electricity34, limit- essential ingredients to the success of virtual learning, is ing its capacity to benefit from remote learning opportu- unequally distributed in LAC, disproportionately affecting nities. In Panama, less than 15 percent of the households poor children in rural areas, and other vulnerable groups in the bottom wealth quintile have access to Internet as such as indigenous children and students with special compared to almost 100 percent of households in the top 33 Rieble-Aubourg and Viteri (2020). 34 WBG et al. (2021). The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response 40 Two Years After | Saving a Generation Figure 1.16. Internet access across LAC, selected countries (%), latest year available Haiti 4% Guatemala 15% Suriname 19% Honduras 20% Dominican Republic 21% El Salvador 23% Peru 24% Total average 24% Guyana 24% Paraguay 25% Mexico 34% Colombia 38% Panama 42% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% Source: Own elaboration based on Barron-Rodriguez et al. (2021a). Note: Data for LAC includes 20 countries including the latest available data for each country. wealth quintile.35 In addition, according to estimates from Figure 1.17. Households in LAC that reported having UNICEF and ITU (2020), about 10 percent of all children problems accessing the Internet (%), mid-2021 in LAC who are enrolled in primary and secondary school 60% (amounting to 13 million students) live in areas where In- 55% ternet cannot be even deployed. 50% Even when households have access to the Internet 40% and electricity, services often have ben intermittent 37% and of poor quality. Available data from the World Bank 30% High-Frequency Phone Survey (HFPS)36 indicate that a sig- 30% nificant share of LAC children living in a household with access to the Internet cannot fully take advantage of on- 20% line learning opportunities. More than half of all families 15% 13% who participated in the survey indicate that the quality of 10% the Internet is poor. In contrast, over one-third indicated that power outages or high-service cost limited their Inter- 0% net access (Figure 1.17). Poor access Power High cost Lack of Lack of quality / outages of internet equipment knowledge speed / data access (e.g., on how to Beyond connectivity, access to devices further impact- packages defective use the devise, internet ed the reach to and participation in online learning. router, etc.) For example, in Colombia, Ecuador and Peru, less than 50 Source: Own elaboration on the basis of LAC’s HFPS, Phase II, Wave 1. percent of all primary students have a computer at home, which points to very low device access rates to effectively 35 UNICEF (2021d). 36 The HFPS is an effort originally led by the Poverty and Equity Global Practice of the WBG. The HFPS collects data on the effects of and responses to COVID-19 across different themes such as labor markets, income, health, gender, food security, education, connectivity, and finance. Detailed microdata for the HFPSs for specific countries or regions can be obtained at https://microdata.worldbank.org/index.php/catalog/hfps/?page=1&ps=15&repo=hfps. In LAC, the first phase was collected in 2020 in 3 waves – May, Jun/July and August - and included 13 countries. The second phase, developed in partnership with the United Nations Develop- ment Program (UNDP), was collected in 2021 in 2 waves – May/July and October/December – and added 11 more countries. The HFPS Phase II, Wave 1 covered an average of 1,150 individuals per country and was nationally representative of all individuals aged 18 or above who have access to a phone in the given country. For further details on the HFPS for LAC, see WBG (2021b). The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response Two Years After | Saving a Generation 41 Figure 1.18a. Share of primary students with a Figure 1.18b. Share of primary students with Internet computer in the household by ethnic origin, selected in the household by ethnic origin, selected countries countries (%) (%) 100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 2016 2019 2015 2019 2015 2017 2015 2019 2015 2019 2016 2019 2015 2019 2015 2017 2015 2019 2015 2019 Brazil Colombia Ecuador Peru Uruguay Brazil Colombia Ecuador Peru Uruguay Afro-descendant Non-Afrod. and non-Ind. Afro-descendant Non-Afrod. and non-Ind. Figure 1.18c. Share of secondary students with Figure 1.18d. Share of secondary students with computer in the household by ethnic origin, selected Internet in the household by ethnic origin, selected countries (%) countries (%) 100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 2016 2019 2015 2019 2015 2017 2015 2019 2015 2019 2016 2019 2015 2019 2015 2017 2015 2019 2015 2019 Brazil Colombia Ecuador Peru Uruguay Brazil Colombia Ecuador Peru Uruguay Afro-descendant Non-Afrod. and non-Ind. Afro-descendant Non-Afrod. and non-Ind. Source: WBG (forthcoming), pp. 31-33. utilize online learning modalities.37 Students from lower Afro-descendant students have access to computers at socio-economic backgrounds, rural areas or specific eth- home, while more than half of white students do.39 nic/indigenous origin were disproportionately affected. In LAC, Afro-descendant students have lower access to a Another important determinant of the reach, participa- computer at home and to Internet in the household, both tion in and quality of the remote learning package was in primary and secondary education, as compared to their the implementation strategy, or the way various deliv- non-Afro-descendant non-indigenous peers (Figures 1.18a ery channels were deployed, which it is difficult to get -1.18d).38 In Brazil, for example, only about one in three right. Whilst many countries developed distance learning 37 WBG (forthcoming). 38 Except in Uruguay, where the implementation of Plan Ceibal and one-computer-per-child has been successful in bridging this gap. 39 WBG (forthcoming). The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response 42 Two Years After | Saving a Generation strategies during school closures in LAC, new studies sug- which connectivity is not a limitation, yet they require an gest that their implementation differed, thereby present- especially strong regular communication stream between ing different challenges and varied effectiveness. Figure households and teachers. Both SMS/phone calls and TV/ 1.19 presents a snapshot of the different strategies fol- radio programs are considered low/medium-tech channels, lowed by Chile, Brazil, Peru, and Uruguay, with varied re- expanding the reach possibility for many students, but also mote learning channels. As can be seen, most countries require robust follow-up and TV/radio programs are not al- have been using multimodal solutions, but they combined ways high quality. For instance, recent evidence from the them differently. For example, Brazil and Peru did not use Dominican Republic, where printed booklets and TV pro- adaptive software for the deployment of their strategy, grams were the most used solutions for distance learning, while Uruguay did not recur to radio programs or print- showed that the usage of the material dropped over time ed take-home materials. Similarly, online sessions in Chile and that significant parental involvement was required to and Uruguay did allow for synchronous learning, but not efficiently utilize the resources provided (Box 1.2).40 On the in Brazil or Peru. On the contrary, while TV programs in other hand, Internet-based strategies, the highest-tech, can Brazil and Peru allowed for interactivity, such feature was potentially propose higher-quality solutions, but are affect- not available in either Chile or Uruguay. ed by both access to and quality of connectivity, as well as by teacher digital capabilities. What is at stake here is both These differences matter because each distance learn- the capacity to combine well different delivery channels ing channel and instrument has its own strengths, lim- and to provide high-quality learning experiences regard- itations, and requirements (see Annex 1.1). For instance, less of the platform (which entails a broader view on the paper and printed material are essentially low-tech tools for quality of the overall response strategy). Figure 1.19. Remote learning implementation strategy in LAC during 2020, selected countries41 Remote Learning Channels Adaptive So ware + Online sessions + TV + Teacher Radio + Teacher Printed Material + Country Teacher Support Teacher Support Support SMS Support SMS Teacher Support Chile Brazil Peru Uruguay The Government implemented the remote learning channel The Government did not implement the remote learning channel Remote learning channel allowed for interactivity Remote learning channel allowed for synchronous learning Remote learning channel allowed for adaptive learning Source: MINEDUC-Chile and WBG (2022). 40 WBG (2022b). 41 This figure was made based on a World Bank study (2022) worked together with the Chilean Education Ministry which compares remote learning implementa- tion strategies through specific characteristics among five delivery channels and five implementation dimensions with a focus on South American countries. None- theless, for a comprehensive review of the multiple platforms implemented by 33 LAC countries, including Central America and the Caribbean, see: https://www. unicef.org/lac/en/online-education-platforms-and-resources . The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response Two Years After | Saving a Generation 43 For effective implementation of any learning modality, Figure 1.20. Disengagement rate, by country (% of the role of the teacher is critical, but despite the efforts student-age population), 2020 vs 2021 made, teacher readiness remained an issue. Evidence is Ecuador indicating that many teachers did not feel prepared for the Bolivia transition to remote learning modalities and overwhelmed Paraguay by the new requirements.42 A teachers’ survey conducted in mid-2020, that included eight countries in LAC, showed Mexico that 57 percent of the respondents agreed that weak tech- El Salvador 15% nological skills were one of their main challenges during LAC 23% the health crisis43.The same survey reflects that 66 percent Honduras of the teachers had completed the digital skills training of- Colombia fered by their own schools, national ministries, internation- Dom. Rep. al organizations, and local non-governmental organizations Argentina (NGOs), but just 56 percent of those trained teachers per- Guatemala ceived that they were ready for effective remote teaching. Costa Rica Further evidence on digital skills is provided in Chapter 3. 0% 20% 40% 60% 2020 2021 Finally, institutional constraints also limited the effec- Source: Own elaboration on the basis of LAC’s HFPS, Phase II, Wave 1. tiveness of the remote learning response. Many coun- tries did not have the institutional capacity to rapidly cope with the pandemic and to adapt to everyday’s new chal- lenges while providing good quality remote learning. Most pandemic, significant in many countries. In Guatemala, ministries of education did not have the capacity to mon- student disengagement was registered at 45 percent in itor the usage of remote delivery systems for education or 2021, which as compared to 17 percent before the pan- students’ engagement and learning. They did not have the demic represents an increase of 156 percent, in Costa Rica technical teams in place and the capacity to continually the increase was 192 percent, from 7 to 21 percent in 2021. monitor and evaluate education processes and outcomes The increasing disengagement in Dominican Republic is to understand if the whole education strategy was being consistent with the very few hours a day of study reported effective. 44 by parents, especially for primary education students (Box 1.2). Based on limited access to connectivity and comput- Emerging regional evidence confirms that remote ed- ers, it is also expected that participation across the region ucation delivery, especially with the limitations and was weaker for the most vulnerable groups (as will also be characteristics above, has fallen short of ensuring par- seen with overall attendance in the next chapter). ticipation. Recent evidence, collected through multiple waves of the HFPS led by the World Bank, indicates that There are also strong indications that the quality of massive school closures and transition to remote school- remote learning has been highly problematic in the ing have had a negative impact on engagement rates in region. First, and in alignment with the institutional the region, measured as participation in any learning ac- constraints highlighted above, according to the latest UN- tivities at home.45 Following the school closures, around 1 ESCO-UNICEF-World Bank-OECD survey on Government of 4 students, who had not left the education system, was responses to the pandemic in the education sector, few not actively engaged in any learning activity by May/July governments have prioritized continuous assessments 2021 (Figure 1.20). The increase of 54 percent in student and evaluations on the effectiveness of remote learning disengagement is alarming, and as compared to before the strategies46, which has constrained the monitoring of these 42 Barron-Rodriguez et al. (2021a). 43 Herrero-Tejada et al. (2020). 44 Barron-Rodriguez et al. (2021a). 45 The “engagement” and “disengagement” indicators come from question 8 of the Education Module in the HFPS, asking if the individual participated in any learning activities at home during the previous week. Percentages (rates) are adjusted for academic break periods. It is worth noting that this indicator of engage- ment does not measure the quality of engagement, i.e. the quality of the work being done. Therefore, it is arguably not possible to fully assess the impact of the pandemic through this metric. 46 This is an issue because this means that important variables to evaluate the effectiveness of remote learning strategies to identify areas of improvement are currently not adequately monitored in the region. See UNESCO, UNICEF, WBG, and OECD (2021a). The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response 44 Two Years After | Saving a Generation Box 1.2. Assessing the effectiveness of remote learning in the Dominican Republic during school closures Governments around the world have designed and implemented different strategies to mitigate learning losses and dropouts caused by the school closures during the pandemic. The Dominican Republic (DR) designed the “Educa- tion for All Safeguarding our Health” strategy. This strategy included different pillars and focused on developing educational materials and tools for students, facilitating connectivity and access to devices for distance learning, and training for teachers and school principals. The strategy also strengthened the administrative data collection to monitor distance learning. The implementation of the strategy was assessed, providing a comprehensive understanding of the functioning and challenges of distance learning in the DR during the pandemic from the perspective of different actors. The report, prepared by the World Bank, uses findings from a nationally representative household telephone survey with 800 households with school age children enrolled in public schools before the pandemic, television ratings for educa- tional programs, as well as a nationally representative phone survey with 454 school directors from public schools. The study found that most students enrolled in public Figure B.1.2.1. Dominican Republic: Proportion schools (84 percent) got access to at least some of the of students that watched distance learning educational materials and tools, all teachers from public programs in Gran Santo Domingo and Santiago schools benefited from a computer to facilitate their con- (% of students aged 4-17) nectivity for distance learning, and most teachers from 70% public schools (89 percent) received training on the use of 62% computers and technologies, socioemotional management 60% of students, and use of communication tools. Despite all 52% 50% these valuable efforts and achievements, data suggests 39% that the performance of distance education faced many 40% 35% 33% challenges like in many other countries. For instance, pri- 30% 24% 25% mary student usage of the TV distance learning programs 30% 32% 28% was 52 percent in November 2020 but dropped to 25 per- 20% 24% 25% cent by April 2021 (Figure B.1.2.1). In addition, the time students viewed the program in April 2021 was between 10% 10-12 percent of the total program duration. 0% Nov Dec Jan Feb Mar Apr 2020 2021 Parental perception indicators collected in that study sug- Pre-Primary & Secondary gests that students could have certain lags when they re- Primary students students turn to face-to-face classes. Less hours of study time per day could mean less learning. Average study hours during Source: WBG (2022b). distance learning have been low for all students. Parents reported that 47% of high school students and 66% of el- ementary students studied on average less than 3 hours a day. With low average hours of study, it is likely that learn- ing was also low. For elementary school, 9% reported that they studied less than an hour per day or nothing at all. Furthermore, 60% both parents of primary and secondary education students agreed in that their children learned less during remote learning than when education was face-to-face. This percentage rises to 84% for primary school principals and 88% for secondary school. Source: WBG (2022b). The COVID-19 pandemic: The heavy toll of the crisis for LAC and the remote learning response Two Years After | Saving a Generation 45 strategies and the measurement of their quality. Second, needed for distance learning; (iii) the complexities in the when there have been assessments, evidence collected in implementation of the specific response strategy; (iv) the a few countries confirmed sub-optimal satisfaction rates. lack of teachers’ preparedness for remote learning; and In Peru, for instance, 67 percent of parents were satisfied (v) institutional constraints. The combination of these fac- with the content and delivery remote learning through TV, tors eventually impinged upon the effectiveness of remote 57 percent through radio and 80 percent through the on- learning. line learning platform Aprendo en Casa (I learn at Home), which shows clear margins for improvement.47 In the Do- Overall, an important lesson derived from the global minican Republic, 60 percent of parents and 88 percent and LAC evidence is the need for a comprehensive ap- of teachers contended that student learning had suffered proach to remote learning, with focus on access, qual- through remote learning as compared to in-person deliv- ity, and governance. The emergency strategies imple- ery. Third, the reopening of schools and emerging learn- mented by various countries were not all that effective to ing assessments are starting to make it eminently clear overcome the challenges of the needed large-scale trans- that this is just the “tip of the iceberg”. As documented in formation. Critical lessons have included the need to use Chapter 3, although Peru and Brazil successfully reached multiple modalities of remote learning in a complementa- 85 and 75 percent of students, respectively, by implement- ry fashion and upskill the capacity of all agents involved ing multimodal delivery systems, even allowing for teach- in the education process (students, teachers and direc- er-student interaction, learning losses are expected to be tors, parents or caregivers, and governments) to provide sizeable. high-quality remote instruction and continuously monitor and evaluate the usage, engagement, and effectiveness of the remote learning strategy for successful real-time decision making. Chile and Uruguay are two examples of 1.4 Key takeaway messages multi-modality approaches to remote learning that have worked through a wise combination of instruments with- The combined impact of the health, economic and ed- in the response package, but also by harvesting on prior ucation crises induced by COVID-19 disproportionately investments in education technologies and pre-existing affected the LAC region. The region ranked atop of the institutional capacity. Peru’s Ministry of Education start- world ranking in terms of the: (i) number of COVID-19 re- ed in April 2020 to supervise the adoption of the national lated deaths – both absolute and per million inhabitants; remote learning program, students’ satisfaction, and the (ii) largest impact on GDP (both during the pandemic and proportion of teachers providing constant feedback to stu- projected into the next two years); and (iii) share of school dents, with encouraging results48. At the same time, co- weeks with schools closed or mostly closed in the last two ordination between key players to mobilize financial and years. This three-fold curse impacted the lives, livelihoods human resources and solve logistical problems is funda- and human capital formation of LAC’s population, in par- mental. 49 ticular that of children and adolescents. However, it is also clear that, even under the best sce- The education systems of LAC made a huge effort to nario, remote learning cannot replace in person educa- cope with the pandemic and ensure learning continuity tion.50 Engagement and quality have been an issue across with remote learning strategies. However, the response the board for remote learning, and the consequences of the faced several limitations due to a multiplicity of factors, combination of school closures and this limited effective- the most prominent being: (i) the uneven state of con- ness will become abundantly clear in the next two chap- nectivity in the region; (ii) the limited access to devices ters when discussing impacts on schooling and learning. 47 Barron-Rodriguez et al. (2021a). 48 Ibid. 49 Muñoz-Najar et al. (2021). 50 WBG (2021a) and WBG et al. (2021). 46 Two Years After | Saving a Generation PHOTO BY: © BISUAL PHOTO / SHUTTERSTOCK.COM Chapter 2 The impact, so far, of the COVID-19 pandemic on schooling The impact, so far, of the COVID-19 pandemic on schooling Two Years After | Saving a Generation 47 T he pandemic generated a silent crisis in the educa- tion sector of the LAC region and only understand- ing the magnitude of the impact could help gauge how 2.1 The state of schooling prior to the pandemic: A positive outlook with good prospects deep it was. As was discussed in Chapter 1, the irruption of the COVID-19 brought about a heavy health and eco- Over decades before the COVID-19 pandemic, compul- nomic toll on countries all over the region. But an equally, sory education attendance had been consistently in- though sadly less loud, crisis was kicked off in the edu- creasing across the LAC region. As Figure 2.1 shows, in cation sector upon the massive shutdown of educational 1989 the out-of-school population (OOSP)51 in LAC com- institutions that took place since. As documented, the in- prised approximately 34.9 million children and adoles- tensity of school closures was such that, on average, the cents. As of 2019, 30 years later, this number had been typical student in LAC missed one full year of in-person more than halved (53% decrease). As a result, prior to education. The impact of the disruption from this unprec- the pandemic outbreak, the estimated OOSP comprised edented exogenous shock permeated all aspects of the around 16.5 million school-aged students. This number educational delivery process, generating a tremendous represented roughly 11% of the total population of com- impact on the schooling and learning process. These next pulsory school age. In other words, on average only 1 in 10 two chapters attempt to understand the breadth, depth school-aged students in LAC did not attend school prior and characteristics of the impact. To place the analysis in to the pandemic. However, even if still large in absolute perspective, the chapters take a two-fold approach. First, terms, the OOSP 30 years earlier was 25%, meaning that they assess the impact on a few educational outcomes, 1 in 4 students of compulsory-education school age did starting by arguably the two most important - schooling not attend school. In sum, half a percentage point of OOSP and learning – and then continue with other outcomes that was being cut yearly before the pandemic outbreak. were put to test in the context of the pandemic. Second, when possible, they carry out an ex-ante and ex-post anal- The reasons behind the progress in school attendance ysis, taking stock of the status of the outcomes prior to the over the decades in LAC were varied. One of the drivers COVID-19 outbreak and after the hit. This chapter focuses was the demographic transition, a worldwide trend that on schooling outcomes. had its own dynamic in LAC, but that led to progressively smaller-sized incoming cohorts into the region’s education Figure 2.1. Enrollment vs. school-age population in Figure 2.2. Population growth rate in LAC (%), LAC (million), 1989-2019 1970 - 2019 160 3.0% 150 2.5% 2.5% 16.5 140 2.2% 130 2.0% 1.9% 34.9 120 1.5% 1.5% 110 1.1% 1.0% 100 0.9% 90 0.5% 80 1989 1999 2009 2019 0.0% 1970 1980 1990 2000 2010 2019 School Age Population Enrollment Source: Own elaboration on the basis of UNESCO Institute for Statistics. 51 The OOSP is measured as the difference between the school-aged population for compulsory levels of education – i.e. the number of children and adolescents of the age range for which compulsory education grades are set in a given school system - and the number of students of the same age range who are enrolled in school at any level of education. The impact, so far, of the COVID-19 pandemic on schooling 48 Two Years After | Saving a Generation Figure 2.3. Death rate index vs. birth rate index in Figure 2.4. School-age population index, LAC (total LAC (per 1,000 people), 1970-2019 and by education level), 1970-2019 1.1 2.0 1.0 1.8 0.9 1.6 0.8 0.7 1.4 0.6 0.6 1.2 0.5 1.0 0.4 0.4 0.3 0.8 1970 1980 1990 2000 2010 2019 1970 1980 1990 2000 2010 2019 Death rate Birth rate Total Pre primary Primary (per 1,000 people) (per 1,000 people) Lower Secondary Upper Secondary Source: Own elaboration on the basis of UNESCO Institute for Statistics. systems with the population annual growth rate falling helped ensure a quicker “catch-up” between the school-aged from 2.5% in 1970 to 0.9% in 2019 (Figure 2.2). A second population and schooling enrollment, as shown previously driver was the dramatic increase in education enrollment, in Figure 2.1, especially starting in the 2000s. As of 2019, especially at the bottom of educational trajectories, with the school-aged population at all levels (pre-primary to up- the sizeable expansion of early childhood education (ECE) per secondary) was declining (Figure 2.4). Arguably then, it coverage, mostly driven by a consolidated approach to- could be assumed that demographics helped ease the finan- wards ensuring universal access to pre-primary education cial burden for school systems to increase coverage rates or, access in the region. The third key driver was the decrease alternatively, that policies aiming to increase schooling at- in drop-out rates across the board in compulsory educa- tendance may have found an “exogenous ally” that may have tion, but most notably in the secondary education levels contributed to reduce the OOSP. (lower and upper). The following paragraphs provide a quick snapshot of each of the three drivers. One of the most important drivers in the reduction of OOSP in LAC was the increase in the compulsory edu- Demographic change in LAC was one of the drivers that cation coverage rates in the region. In fact, as of 2019, contributed to the progressive reduction in the OOSP. The prior to the pandemic outbreak, around 134 million chil- strong demographic change that took place in LAC in the last dren and adolescents of compulsory education age were 50 years worked as an exogenous change that, other fac- enrolled in school52. As hinted above, this meant that tors constant, helped reduce pressure on education systems roughly 9 in 10 students who should be attending school, across the region by decreasing the size of incoming cohorts. given their age, were already in the system (Figure 2.5). As In fact, the birth rate in LAC decreased more than 50% in shown above, increased enrollments, coupled with a de- the last 50 years (Figure 2.3) while the death rate did not clining school-age population, led to increasing coverage fall more than 40%. This significant change in demographics rates53. Hence, right before the arrival of the pandemic, the 52 This report refers to compulsory education levels of education to the four levels of education (pre-primary, primary, lower secondary and upper secondary) for which at least one country in LAC has confirmed the full given level as compulsory by law. In fact, compulsory education laws in LAC present significant differences across countries because of at least the four following reasons: (i) different levels of education have a different number of years of education (e.g. primary educa- tion could last 6 or 7 years); (ii) different levels of education have different starting age (6 years of age or 7 years of age); (iii) some countries do not set compulsory education by levels of education, but by age groups covered within a given level (e.g. the kindergarten year for pre-primary); and (iv) age groups are defined differ- ently in different laws (e.g. the last year of upper secondary education could be for 17-year-olds or for 18-year-olds or for a combination of the two). For a compre- hensive review of the age groups for which compulsory education laws are set for each LAC country, see Annex 2.1. 53 All pre-primary indicators presented in this section are calculated based on gross enrollment while indicators for primary, lower secondary, and upper second- ary are based on net enrollment. Gross enrollment rate is defined as the number of students enrolled for a given level of education, regardless of age, expressed as a percentage of the official school-age population corresponding to the same level of education. The impact, so far, of the COVID-19 pandemic on schooling Two Years After | Saving a Generation 49 Figure 2.5. Total enrollment in LAC, by level of Figure 2.6. Net enrollment rate in LAC, by level of education (million), 2000-2019 education (%), 2000-2019 80 100% 97.2% 93.4% 90% 60 56.6 80% 77.8% 78.7% 40 32.4 70% 23.9 21.5 20 60% 0 50% Pre-Primary Primary Lower Upper Pre-Primary Primary Lower Upper Secondary Secondary Secondary Secondary 2000 2005 2010 2015 2019 2000 2005 2010 2015 2019 Source: Own elaboration on the basis of UNESCO Institute for Statistics. primary and lower secondary education levels presented attending school, but also improved internal efficiency – almost universal coverage, with net enrollment rates of with a lower share of the age group attending lower levels about 97% and 93%, respectively. At opposite extremes of of education, i.e. primary or lower secondary education the compulsory education trajectory, though, pre-prima- (LSE), due to repetition or overage problems (Figure 2.8). ry and upper secondary education levels presented lower The authors hypothesized that two factors have apparent- coverage rates, reaching 78% and 79%, correspondingly. ly contributed to lowering drop-out rates over time in the Of importance, by 2019, the net enrollment rate for LAC region: (i) increased returns to education over time; and had increased with respect to 2000 at every level of com- (ii) an “income effect”55. Increasing returns to education pulsory education, with effects most noticeable in pre-pri- over time have also been a reason identified by Bassi et al. mary (with a sizeable increase, especially between 2005 (2015) to explain the improvements in enrollments as well and 2010) and in upper secondary education (Figure 2.6). as graduation rates and decrease in dropouts. Other expla- nations also pointed out by these authors included more Other important drivers in the reduction of OOSP in LAC resources allocated to education and the implementation were the decrease in drop-out rates in compulsory edu- of policies to help students stay in school, including wel- cation, most notably at the lower and upper secondary fare programs increasing the demand for schooling. education levels, and the increase in graduation rates. As Kattan and Székely (2017) document in their cohort As a result of these trends, before the COVID-19 pan- analysis of school dropout in the LAC region54, significant demic, the OOSP, both in absolute terms and as a share progress was achieved since 1979 or so. In fact, while the of the school age population, was consistently decreas- cohort of LAC students born in the 1979-1981 period had, ing over time across education levels. As of 2019, from on average, only 1 in 2 students enrolled in upper second- the 16.5 million school-aged children and adolescents that ary education (USE), the cohort of students born 15 years were out of school, 6.1 million were out of pre-primary later (in the 1994-1996 period) had 7 students out of 10 education, 1.6 million out of primary, 2.3 million out of attending USE (Figure 2.7). In addition, not only did later LSE, and 6.5 million out of USE (Figure 2.9). All of these cohorts show a lower share of USE-aged adolescents not values, with the exception of those for LSE, were at their 54 The paper, which draws from a series of household surveys, only covers 18 LAC countries, including 17 Spanish-speaking countries and Brazil. 55 The increase in the returns to education over time seems to have pulled an ever-increasing number of people of compulsory education age to enroll in school. This factor in particular appears to have played a prominent role in increasing graduation rates for lower secondary education, thereby contributing to positively impact enrollment levels at USE. The “income effect” (which describes the change in the quantity demanded for a good given a change in price that affected real income) seems to have stemmed from a more favorable macroeconomic and labor market environment, especially during the 21st century, where inflation rates in the region tended to decline sharply, leading to higher retention rates due to decreased direct and indirect costs of schooling. The impact, so far, of the COVID-19 pandemic on schooling 50 Two Years After | Saving a Generation Figure 2.7. Evolution of enrollment in LAC for 4 Figure 2.8. Distribution of upper secondary education cohorts, 1990-2010 (% enrolled) (USE)-age students in LAC by schooling status for 4 cohorts entering USE between 1994 and 2011 (%) 100% 94% 96% 94% 94% 95% 93% 90% 90% 100% 90% 86% % of USE age youth by schooling status 83% 90% 80% 29% 27% 25% 77% 37% 80% 72% 70% 70% 70% 68% 25% 61% 60% 25% 60% 26% 56% 50% 27% 50% 52% 40% 46% 40% 30% 39% 48% 51% 45% 33% 20% 30% 35% 28% 10% 20% 25% 20% 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 0% Cohort 1 Cohort 2 Cohort 3 Cohort 4 Cohort 1: Born in 1979-81 Entering primary (entering (entering (entering (entering in 1985-87 and entering USE in 1994-96 USE in USE in USE in USE in 1994-96) 1999-01) 2004-06) 2009-11) Cohort 2: Born in 1984-86 Entering Primary in 1990-92 and entering USE in 1999-01 % of USE age youth attending USE Cohort 3: Born in 1989-91 Entering Primary in 1995-97 and entering USE in 2004-06 % of USE age youth not enrolled in school % of USE age youth attending LSE or Primary Cohort 4: Born in 1994-96 Entering Primary in 2000-02 and entering USE in 2009-11 Source: Kattan and Székely (2017), Figure 1. Source: Kattan and Székely (2017), Figure 2. minimum historical values ever. Interestingly, while in ab- gender gaps, the OOSP showed a slight gender bias against solute terms the USE level of education had the largest ab- boys, for every education level and both in absolute and solute number of OOSP, due to the demographic transition relative numbers (Figures 2.11 and 2.12). As regards OOSP’s explained above, the pre-primary education level had the differences according to socio-economic status (SES), low- largest share of OOSP (22.2% versus 21.3% for USE, Figure er-income quintiles presented lower attendance rates than 2.10). In a nutshell, before the pandemic, 3 out of 4 OOSP higher-income ones. For example, in Mexico, the probabili- were located at the opposite extremes of the distribution ty of an adolescent aged 15-17 being OOSP was 30% higher of the compulsory education cycle, namely pre-primary if the student belonged to the lowest quintile of the distri- and USE. Such result was partially explained by legal dif- bution of income rather than to the top one57. ferences across countries in the definition of the compul- sory education age.56 Finally, the OOSP average situation for LAC prior to the 2.2 The COVID-19 pandemic shock pandemic masked significant differences across several and its impact on schooling dimensions, as it would be expected in such an unequal region. In terms of countries, for example, the Dominican The arrival of the COVID-19 pandemic in LAC entailed a Republic’s OOSP for pre-primary education was 9 times significant disruption to school attendance, raising ex- higher than Brazil’s (43% versus 5%). When it comes to pectations that the downward trend in the OOSP would 56 For example, only 36 countries out of 45 in the region (80%) have established a compulsory education upper bound at 15 years of age. Similarly, only 38 of the 45 countries in LAC (84%) have established at least one compulsory year of pre-primary education. For further details, see Annex 2.1. 57 Estimates based on 2018 household survey available for Mexico at the Socio-Economic Database for Latin America and the Caribbean (SEDLAC). See https:// www.cedlas.econo.unlp.edu.ar/wp/en/estadisticas/sedlac/. The impact, so far, of the COVID-19 pandemic on schooling Two Years After | Saving a Generation 51 Figure 2.9. OOSP in LAC, by compulsory education Figure 2.10. OOSP in LAC, as a share of school-aged level and for selected years (million), 2000-2019 population, by compulsory education level and for selected years (%), 2000-2019 30 80 60 20 Percent Million 40 10 22.2% 21.3% 6.1 6.5 20 1.6 2.3 6.6% 2.8% 0 0 Pre-Primary Primary Lower Upper Pre-Primary Primary Lower Upper Secondary Secondary Secondary Secondary 2000 2005 2010 2015 2019 2000 2005 2010 2015 2019 Figure 2.11. OOSP in LAC, by gender and compulsory Figure 2.12. OOSP in LAC, as a share of school-aged education level (thousands), 2019 population, by gender and compulsory education level (%), 2019 4,000 500 25% 21.5% 30% 3,500 3,463 450 22.8% 22.4% 3,223 400 20.1% 25% 2,992 20% 3,000 2,905 350 Percentage di erence 2,500 20% 300 15% 11.6% Thousand 2,000 250 15% 10.3% 1,500 200 10% 1,303 150 7.3% 10% 1,000 930 977 5.8% 706 100 5% 3.1% 5% 500 2.5% 50 0 0 0% 0% Pre-Primary Primary Lower Upper Pre-Primary Primary Lower Upper Total Secondary Secondary Secondary Secondary (all levels Female Male Di erence (bias against boys) of education) Female Male Percentage Di erence (bias against boys) Source: Own elaboration on the basis of UNESCO Institute for Statistics. be interrupted. Theoretically, the historical OOSP trend OOSP) (Figure 2.13). The initial assessments of COVID-19 could either: (i) continue decreasing (either by not being effects pointed to the latter as the most likely scenario, influenced at all by the shock or by decreasing at a slow- with a projected increase in OOSP from March 2020 on- er pace); (ii) stagnate (i.e. stopping the descent, but not wards58. The initial estimates of the impact of the pandem- deteriorating); or (iii) reverse (through an increase in the ic on school drop-out pointed to an estimated 7 million 58 The transmission mechanisms that were forecasted (UNESCO, 2020a) were, for example, student disengagement (as a result of a forced remote learning), the need to support household work, income loss (which could lead to child/adolescent labor, an increased opportunity cost of staying in school), and other problems associated with lockdowns (e.g. early marriage, teenage pregnancy). In addition, since the pandemic interrupted critical stages of the educational process at all ed- ucational levels, it was also foreseen that such interruptions would lead to higher dropout rates (Azevedo et al. 2020). The impact, so far, of the COVID-19 pandemic on schooling 52 Two Years After | Saving a Generation Figure 2.13. Possible trajectories of the OOSP in LAC Figure 2.14. Students at risk of dropping out in LAC, after the COVID-19 pandemic (million) by education level (thousands and %) 22 700 35% 600 30% 18 500 25% Thousand students 400 20% Percent 14 300 15% 200 10% 10 2010 2015 2019 2024 100 5% Reverts Stops downward Continues pre-COVID trajectory, pre-COVID 0 0% Pre-primary Primary Lower Upper trajectory: halting progress, decreasing Secondary Secondary Increasing but not trajectory OOSP deteriorating # at-risk students % at risk students Source: Own elaboration. Source: UNESCO (2020). students around the world leaving school permanently any point before the school year ends; and (iii) at con- due to the COVID-19 shock59. Specifically for LAC, UNESCO clusion – i.e. after passing the given school grade, by not (2020) estimated that 2 million students were at risk of enrolling in the immediately following one. In the para- dropping out, with the pre-primary and USE levels expect- graphs below, the impact of the pandemic on the OOSP is ed to be the most affected (Figure 2.14), and students from explored from these three angles, though focusing more the most disadvantaged backgrounds being the most dis- on attendance rates, which were the ones immediately proportionately affected, especially in low- and lower-mid- impacted upon the pandemic outbreak60. Throughout the dle income countries. analysis, we draw from three main sources of data: coun- try-specific administrative data from Education Manage- Regrettably, actual data that could help document the ment Information Systems (EMISs), the HFPS (described impact of the pandemic on schooling is still scarce at in Chapter 1), and country-specific National Household this stage. Nonetheless, results from surveys especial- Surveys. ly designed to understand the impact of the pandemic, National Household Surveys from different countries, In terms of enrollment, detailed administrative data and administrative data are starting to give some infor- already available for a few countries for 2021, seemed mation about the direction and magnitude of the effects to show at that stage no clear impact of the pandem- of COVID-19 on schooling and, as a result, on OOSP. In ic two years after, on average. In fact, and as part of fact, the OOSP phenomenon can be understood as one an ongoing Human Capital Review (HCR)61 for Central that could happen at three different points/stages of the America62 conducted by the World Bank, data from the school year: (i) at enrollment – for students who enroll Dominican Republic, Guatemala, and Panama appear to at school, but then disenroll or do not attend dropping carry two conclusions: (i) for those countries where en- out before the school year begins; (ii) at attendance – at rollment slightly decreased from 2019 through 2021, it is 59 Azevedo et al. (2020). 60 In fact, as was explained in Chapter 1, the timing of the pandemic (early to mid-March for most of the countries in LAC) caught the vast majority of the school systems in the region with an already advanced school year or with a school year which had recently kicked-off. See Section 1.2 and Figure 1.6. 61 The HCR is a new core diagnostic tool launched by the World Bank in early 2020. The HCR is an analytical product that uses new or existing data and accept- ed analytical methods to: (i) present an overview of the state of key human capital endowments and challenges in a country; (ii) identify priority human capital outcomes that require improvement; (iii) offer country/region-specific policy options to improve these outcomes; and (iv) identify areas for further analysis in key sectors. HCRs are being conducted in a handful of countries around the world and, specifically, for LAC, the ones for Brazil, Paraguay, and the 6 countries in Central America are expected to be completed between 2022 and 2023. 62 The Dominican Republic, though technically a Caribbean country, is considered part of Central America within the current geographical organization of the Bank for convenience purposes – having the same language (Spanish) as the 5 other countries located in the Central American isthmus (Costa Rica, El Salvador, Guatemala, Honduras, and Panama). The impact, so far, of the COVID-19 pandemic on schooling Two Years After | Saving a Generation 53 Figure 2.15. Dominican Republic: Total enrollment Figure 2.16. Panama: Total enrollment index, by index, by level of education, 2019-2021 school age group, 2019-2021 1.02 1.20 1.01 1.15 1.14 1.00 0.99 0.99 1.10 0.99 0.98 0.98 1.05 1.04 0.97 1.02 1.00 0.96 0.95 0.95 0.94 0.90 0.93 2019 2020 2021 2019 2020 2021 Total Primary Secondary 6 to 14 15 to 17 6 to 17 Source: Own elaboration based on administrative data at school level for Source: Own elaboration based on administrative data from Panama’s SIDE school years 2019-2020, 2020-2021 and 2021-2022. (Integrated Statistical Data System). not possible to disentangle the potential “pandemic ef- Moving now to attendance and using the latest avail- fect” from the “demographic transition” effect described able information from the HFPS64, survey data pointed above; and (ii) there seems to be a consistent “push-and- to significant reductions in attendance rates when com- pull” effect with students moving from private schools to paring to the pre-pandemic situation by mid-202165. In public schools, as evidenced by the sustained increase fact, as shown in Figure 2.17, attendance rates66 as of mid- in enrollment in public schools vis-à-vis the steady (and 2021 were 12% lower for the region than pre-pandemic sometimes very significant) decrease in enrollment in attendance rates (February 2020). As with the case for private schools. Regarding the first point, administrative almost every education indicator, there were important data from the three countries listed above show an over- differences across the board. While countries like Guate- all neutral effect of the pandemic on total enrollment in mala, Honduras, and Jamaica presented significant per- compulsory education63. In fact, while Panama presents centage differences between May/July 2021 and February a 4% increase in total enrollment, Guatemala and the 2020 (33%, 22%, and 20%, respectively), countries such Dominican Republic displayed mild drops of 2% and 1%, as Costa Rica and St. Lucia show much smaller percentag- respectively (Figures 2.15 and 2.16). Such variations, es- es (10% and 6%). Chile only experienced a 1% decrease pecially the latter ones, cannot really tell apart how much in attendance. While many or most schools were still of the variation could be genuinely attributed to the pan- closed by mid-2021 in the region, these different trends demic at this stage, as there is an ongoing demograph- seem consistent with different paces in school reopening ic transition underway for the region, which may play a and the reach of the remote learning response, with Chile role. The second stylized fact (the “push-and-pull” effect) and Costa Rica doing better on both counts than countries is more consistent across countries, even if the causes like Guyana, Guatemala, and Honduras. When disaggre- behind this effect are yet to be investigated with the help gating by age groups (Figure 2.18), we see that the per- of microdata. A more in-depth analysis of this issue is un- cent difference between the before- and during-pandemic dertaken in Box 2.1. attendance rate is significant for both students between 63 Compulsory education, in this case, only refers to the consolidated information from the primary and secondary education levels but does not include pre-pri- mary education. 64 The 2nd Phase of the HFPS was originally planned to be carried out towards the end of 2021. Unfortunately, data collection continued, for some countries, through January 2022. Hence, as of the date of the release of this report, it was not possible to analyze the microdata from this wave. 65 The bulk of the HFPS collection period took place between May and July 2021. 66 “Attendance rate” was defined, in the HFPS, as the proportion of school-age population that had done any in-person (attended classes face-to-face) or remote educational activity (participated in learning activities at home) at the time of the survey. Percentages (rates) are adjusted for academic break periods. The impact, so far, of the COVID-19 pandemic on schooling 54 Two Years After | Saving a Generation Figure 2.17. Attendance rates before and during the Figure 2.18. Attendance rates before and during pandemic, for selected countries (%), the pandemic, by age group for selected countries February 2020 vs. May/July 2021 (% change), February 2020 vs. May/July 2021 0% Chile 97% 98% St Lucia -2.3% Costa Rica -5% Brazil LAC 86% 97% -8.4% Dom. Rep. -10% -11.0% Haiti -12% Colombia -13.5% -14.0% -15% Mexico -15.8% -16.6% Jamaica -17.3% -18.5% Honduras -19.3% -20% Belize Guatemala Guyana 64% 97% -25% [15-17] [15-17] [15-17] [15-17] [15-17] [6-14] [6-14] [6-14] [6-14] [6-14] Total 0% 10% 20% 30% 40% 50% 60% 70% 80% 90%100% During pandemic Before pandemic LAC LAC Argentina Brazil Colombia Mexico Source: Own elaboration on the basis of LAC’s HFPS, Phase II, Wave 1. Source: Own elaboration on the basis of LAC’s HFPS, Phase II, Wave 1. 6 to 14 and 15 to 17 years old, being a bit higher for the portray trends in OOSP as a share of school-age popula- latter (11% vs. 15.8%). A similar trend is consistently ob- tion across four countries68. Three of the four household served across countries in the region. It is important to surveys took place at the end of 2021 when schools in the understand that the comparison between before and after respective countries had largely reopened, another one the start of the pandemic is done between two different (Costa Rica) took place at the same time as the HFPS, when months/time periods each year so the attendance indica- schools were already fully open.69 For analytical purposes, tor could be biased (given the seasonality of this indica- data trends in OOSP70 are presented for three school years tor). Notwithstanding this and other HPFS’ limitations67, - the pre-pandemic (PP) year (2019), the “in-pandemic, no findings from the HFPS II (Wave 1) help in having a better vaccines yet (IPNV)” year, and the “in-pandemic, vaccina- understanding of the effects of the pandemic on educa- tion available (IPVA)” year – and for two age group break- tion systems across LAC, with the dramatic consequences downs – the 6-to-14-year-olds, which is the group for which that decreased attendance can have on drop-out rates and OOSP only represented a minor percentage before the pan- learning gaps later on. demic, and the 15-to-17-year-olds, which is the group with the largest absolute number of people in the OOSP. Three National Household Surveys for selected countries show important conclusions seem to emerge: (i) the average a somewhat different picture. Figures 2.19 to 2.22 below downward OOSP trend, calculated as a ratio of school-age 67 The main two limitations are related to the representativeness of the sample. Country representativeness was limited to the 18+-year-old population with ac- cess to a phone. There was an extra limitation: HPFS the Education Module was applied but to only one (randomly chosen) school-age person in the household. 68 Given the necessary time and data availability limitations of processing microdata from 45 different country-specific household surveys in the region, this section draws on a sample of 4 selected countries that represent very large countries (Brazil), medium-sized ones (Argentina and Colombia) and small-sized ones (Costa Rica) for which comparable data as of late 2021 are available. 69 Household Survey data presented correspond to: (i) the third trimester (2019, 2020, 2021) for Argentina; (ii) the fourth trimester (2019, 2020, 2021) for Bra- zil; (iii) the month of November (2019, 2020, 2021) for Colombia; and (iv) the month of July (2019, 2020, 2021) for Costa Rica. 70 For this analysis, since data are not drawn from administrative records from the Ministries of Education, the OOSP is calculated as the difference between the school-age population in the 6-17 years of age group (i.e. not including any age group below the primary education level) and the population attending school in the corresponding age group (6-17). The impact, so far, of the COVID-19 pandemic on schooling Two Years After | Saving a Generation 55 Figure 2.19. Argentina: OOSP as a share of the Figure 2.20. Brazil: OOSP as a share of the corresponding school-aged population, total and by corresponding school-aged population, total and by age group (%), 2019-2021 age group (%), 2019-2021 10% 25% 9% 8% 8.1% 20% 19.4% 7% 6% 15% 13.0% 5% 4% 4.2% 10% 3% 2.7% 5.8% 2% 2.1% 5% 4.2% 1.4% 1% 0.9% 1.0% 1.2% 0% 2019 2020 2021 0% 2019 2020 2021 6 to 14 years 15 to 17 years 6 to 17 years 6 to 14 years 15 to 17 years 6 to 17 years Source: Permanent Household Survey,third quarter. INDEC. Source: PNAD, fourth quarter. Figure 2.21. Colombia: OOSP as a share of the Figure 2.22. Costa Rica: OOSP as a share of the corresponding school-aged population, total and by corresponding school-aged population, total and by age group (%), 2019-2021 age group (%), 2019-2021 30% 9% 8% 7.8% 25% 24.2% 7% 20% 18.6% 6% 5% 4.5% 15% 4% 3.4% 10% 9.2% 3% 8.0% 2.3% 2% 5% 4.6% 4.9% 1% 0.6% 0.8% 0% 0% 2019 2020 2021 2019 2020 2021 6 to 14 years 15 to 17 years 6 to 17 years 5 to 12 13 to 17 5 to 17 Source: National Household Survey (GEIH), November. DANE. Source: National Household Survey, July. INEC-Costa Rica. population and using attendance rates, in the PP year ap- experienced a small cumulative absolute increase in 6-14 pears to have continued decreasing after two years since OOSP of about 97 thousand children representing a 15% the pandemic hit; (ii) the 15-17 age group in OOSP consis- increase vis-à-vis the pre-pandemic. On the other hand, no tently - and for the most part, significantly - decreased to- such overall increase in absolute numbers is visible for the wards the IPVA year – with a “kink” in Brazil and Colombia 15-17 OOSP for the countries analyzed. To sum-up, there is in 2021 inflecting somewhat the trend; and (iii) the 6-14 some evidence that OOSP in selected LAC countries, and in age group in OOSP consistently – although less significant- proportion of school-age population, seems to have contin- ly – increased towards the IPVA year. The four countries ued decreasing over time, with a generally decisive drop in The impact, so far, of the COVID-19 pandemic on schooling 56 Two Years After | Saving a Generation the 15-17 age group, which made up for the less significant above, and which could affect more this age range, it is though positive increase in the OOSP in the 6-14 age group. possible that the attendance rate of this cohort by the end The decreasing trend in the OOSP 6-14 age group has how- of 2021 reflects the prioritization in school reopening. ever been reversed for these countries. Specific policies, benefitting all students, but especially the 15-17 years old, may also explain the enrollment and Attendance rates for the 6-14 years old show at least attendance trends in 2021. Explanations for the trends some decrease. It is difficult to compare phone and na- may be related to automatic promotion policies (which tional surveys because they may capture different dimen- were for instance encouraged in Argentina, Brazil, El Sal- sions, and over- or under-estimate the extent of the prob- vador, and the Dominican Republic), and cash transfer lem71, but at least in the case of the 6-14 years old, while policies implemented notably in Colombia and Brazil.73 different surveys show different magnitudes, they point for Outcomes would have been worse without these policies. the selected countries to decreases in attendance over the In Brazil, increased age-for-grade distortion could also be pandemic period. Different magnitudes may also be relat- part of the explanation. According to Todos pela Educação ed to the different timing of the surveys, insofar as Colom- (2021), one of the reasons that explains that attendance bia, Argentina and Brazil had experienced larger school re- continued increasing after the pandemic is the increase opening by the time of their national surveys (most of the in the number of students attending school but in previ- schools were open in those countries by then), which may ous years that do not correspond to their age (increase in have helped boost attendance72. Nonetheless, the negative age-for-grade distortion). In fact, the number of students trends point to some concerns with the capacity of these in this situation was around 1.5 million in 2019, 1.7 million countries to have kept children engaged and attached to in 2020, and 1.6 million in 2021, more than in 2019. Final- school. As schools were closed during the pandemic, re- ly, it is also simply possible that direct and indirect costs mote learning was the only alternative offered by most as well as opportunity costs of attending school may have schools which, as discussed previously, had many lim- in fact decreased following the availability of virtual edu- itations and challenges. In many cases, for younger kids, cation and the difficulty of finding good jobs in the middle studying remotely implied needing support and assistance of the crisis (opportunity costs, however, may have played from their parents who not always were able to assist out differently across countries) keeping youth in school. them. In other words, younger kids were more dependent While these are positive findings, the inflection points in on a third person to attend school remotely. Moreover, the the 15-17 years old OOSP in Brazil and Colombia in 2021 limited effectiveness of remote schooling may have espe- may nonetheless be seen as a signal that this group will cially affected the engagement and attendance of that age continue needing close attention. cohort. This may have led to higher risks of disconnect from school, made worse by the prioritization of higher While the full magnitude of the impact of the pandem- grades in school reopening. ic on schooling is not clear yet, results from the HPFS show that by mid-2021: (i) the most affected by the Data are less consistent for the 15-17 years old, where pandemic in terms of attendance rates were lower-in- national surveys are showing continued increases in come children and youth, but (ii) the gender gap had attendance rates comparing with the pre-pandemic, decreased for children in households with access to one which could also simply indicate no clear impact (yet) or less assets. A couple of national household surveys on this age range by the end of 2021. Beyond method- show more nuanced outcomes. Using the HPFS II, wave ological differences between surveys already pointed out 1, we found that the impact on attendance rates was much 71 First of all, the definition of the attendance rate indicators is different in each source –in the HPFS, the attendance rate is capturing the proportion of school- age population that had done any in-person (attended classes face-to-face) or remote educational activity (participated in learning activities at home) at the time of the survey for the pandemic period, while the attendance rate in the household survey measures the proportion of school-age population who reported to be at- tending any educational institution during the period of reference. Secondly, the cross-sectional nature of national surveys may be overestimating attendance rates in the pandemic period. And it is also simply possible that the “attendance rate” is being over-estimated in areas and/or periods with many schools closed (risk mit- igated by the timing of the surveys), with families responding positively even in cases of limited “engagement”, which would lead to a convergence with the phone surveys, possibly more attuned to capturing the realities on the ground. The measured attendance rates may capture more of a “formal” attachment rather than actual attendance. On the other hand, design and implementation aspects of the HPFS, especially the possible impact of seasonality on the indicators, may impact the reliability of their results and possibly over-estimate some impacts. 72 On the other hand, Costa Rica is showing a somewhat smaller difference between survey types, possibly also because they took place at the same time. 73 The largest share of federal resources to combat COVID-19 was to strengthen cash transfers during the COVID-19 crisis. In 2021, Bolsa Família (PBF), the re- nowned conditional cash transfer program in Brazil, was expanded to include an additional 1.2 million new families from the program’s waiting list. A recent anal- ysis shows that an increase of 10 percent of the number of families who receive PBF would explain a 0.3 percent increase in years of schooling in alignment with other literature (WBG, 2022a). The impact, so far, of the COVID-19 pandemic on schooling Two Years After | Saving a Generation 57 Box 2.1. Dropouts in private but not in public schools: A push-and-pull effect? Detailed administrative data for select countries show a consistent pattern that took place between 2019 and 2021: even though total enrollment does not seem to have been affected much by the pandemic, yet, the distribution of enrollment across institutions within the system has. Indeed, while public schools experienced an increase in enroll- ment, private schools showed a (significant) decrease in enrollment. Data from the Dominican Republic (DR), Guate- mala, and Panama all confirm the same stylized fact. Figure B.2.1.1. Dominican Republic: Enrollment, Figure B.2.1.2. Guatemala: Enrollment, 2019-2021 2019-2021 1,000,000 10% 2,500,000 3% 5% 4% 900,000 2% 0% 0% 800,000 2,000,000 700,000 -5% -13% -10% -5% 600,000 1,500,000 -7% -10% 500,000 -23% -20% 400,000 -15% 1,000,000 -15% -30% 300,000 -20% 200,000 500,000 -40% -25% 100,000 0 -50% -30% 0 6 to 11 12 to 17 6 to 11 12 to 17 7 to 12 13 to 17 7 to 12 13 to 17 years years years years years years years years Public Private Public Private 2019 2020 2021 2019 2020 2021 Percentage Di erence 2021 vs. 2019 Percentage Di erence 2021 vs. 2019 Source: Own elaboration based on administrative data. Source: Own elaboration based on administrative data. The specific pattern of changes in enrollment during Figure B.2.1.3. Panama: Enrollment, 2019-2021 the pandemic lead to the immediate question: was 700,000 40% there a push-and-pull effect with students who were formerly in private schools moving to public ones? But 600,000 19% 20% if so, in those countries where there was a net decrease 500,000 5% in total enrollment (e.g. DR or Guatemala), was there not a net loss of students, even after factoring in de- -4% 0% 400,000 mographics? If so, who were these students, those that -17% were at higher risk of dropping out in public schools, 300,000 -20% but none of the ones that came from private schools? 200,000 To be able to properly respond to these questions, it is -40% fundamental to get access to the microdata to follow 100,000 the same individuals over time (panel data on the co- hort of students still in school in March 2020). The Hu- 0 -60% 6 to 14 15 to 17 6 to 14 15 to 17 man Capital Review being prepared by the World Bank years years years years during 2022 should help shed light on this. Public Private 2019 2020 2021 Percentage Di erence 2021 vs. 2019 Source: Own elaboration based on administrative data. (Box continues on next page) The impact, so far, of the COVID-19 pandemic on schooling 58 Two Years After | Saving a Generation Box 2.1. Dropouts in private but not in public schools: A push-and-pull effect? (continued) Now, there is another question: was the significant increase in public school enrollment just an artifact of the “push- and-pull” between private and public schools? Or was there an important “retention effect” that discouraged drop- ping out of school for those students who were already in public schools? Again, the lack of microdata at this stage precludes a more conclusive analysis. Nonetheless, it is worth mentioning that the Central American countries un- der analysis made substantial efforts to ensure schooling continuity during school closures. Providing students and teachers with digital devices in the Dominican Republic, the activation of the accelerated learning program in Pan- ama, or the school insurance and the continuity of the school feeding programs in Guatemala, were all policies that may have not only prevented students in public schools from dropping out, but that could have even incentivized students in low-cost private school to feel attracted to enroll in public schools in an effort to help lower both direct and indirect costs of schooling for their families. In any case, despite the results gotten so far, and the observed trend in enrollment in these countries in Central Amer- ica, future developments are uncertain. All efforts and new policies implemented in the countries to counteract the negative effects of the COVID-19 pandemic seem to have had positive implications in school enrollment, specifically for children and youth in public schools. In the future, however, this trend may be reverted when all the “emergency” policies are over, and the students get fewer incentives to enroll as well as when opportunity costs of education in- crease due to better economic conditions and a more dynamic labor market. Source: Own elaboration based on administrative data for the DR, Guatemala, and Panama. higher for those children and youth in families with access effect of automatic promotion policies. These findings also to 1 or fewer assets than for those of families with access seem consistent with the possible “retention effect” in to 3 or more assets (Figure 2.23). The difference in atten- public schools in Central America. More data and surveys dance rates between these families is particularly strong are needed to assess the full impact of the pandemic on for the 6-14 years old. In addition, according to the survey, lower-income children and youth. the gender gap in attendance rates increased from 0.9% before the pandemic (February 2020) to 2.3% during the pandemic (May/July 2021). Nonetheless, the gender gap was reduced for children belonging to households with ac- 2.3 Key takeaway messages cess to one or less assets while it increased for those with access to two assets and in a greater proportion for chil- The relatively limited available evidence so far on the dren in families with access to three or more assets (gap impact of the pandemic on schooling likely points to increased from 0.5% to 3.6%) as shown in Figure 2.24. an important attendance shock during remote learning Data from the national survey of Colombia show similar for all children and youth, but also to an overall limit- trends for the gender gap but somewhat different trends by ed impact on schooling by the end of 2021 thanks to a income level. In fact, the attendance rate of lower-income set of protective policies. Before the pandemic, the re- 15-17 years old increased more than for their higher-in- gion was on a positive trend towards continuing reducing come counterparts. Such differences could be explained its 11% OOSP rate at a 0.5 percentage-point pace per an- by a combination of successful policies implemented by num, by the time the pandemic hit. The massive closure the government during the pandemic to protect lower-in- of educational institutions led many education research- come youth, including the continuity of school feeding pro- ers and practitioners worried about a sudden reversal in grams, low-fee policies, and possibly more difficult access this trend. Available information is showing that there of lower-income youth to job opportunities. The national was most likely a negative shock to schooling attendance survey data for Argentina also seem to show slightly higher when schools were mostly closed across countries, espe- relative increases in the attendance rate of lower-income cially strong for lower-income students. Yet, by the end of vis-à-vis higher-income youth, pointing to the equalizing 2021, when schools were mostly open and looking at other The impact, so far, of the COVID-19 pandemic on schooling Two Years After | Saving a Generation 59 Figure 2.23. LAC: Attendance rates before and during Figure 2.24. LAC: Gender gaps in attendance rates the pandemic, by age group and access to assets (% before and during the pandemic, by access to difference) assets (%) 0% 0% -1% -0.5% -6% -0.7% -0.6% -1% -10% -13% -2% -1.4% -18% -19% -2% -22% -2.1% -3% -3% -4% -3.6% -40% -4% 0 - 1 assets 2 assets 3+ assets 6 to 14 15 to 17 6 to 14 15 to 17 6 to 14 15 to 17 0-1 assets 2 assets 3+ assets Before Pandemic During Pandemic Source: HPFS II, Wave 1. Source: HPFS II, Wave 1. indicators and selected countries, the overall impact on start monitoring. At the same time, continuous attention is attendance/enrollments rates appeared limited, especial- also warranted to the 15 to 17 years old, especially as they ly for the 15 to 17 years old. Reduced opportunity costs represent the larger OOSP group. of schooling, cash transfer policies, automatic promotion rates, among other factors, may have played a role in this Future follow-up surveys and administrative data outcome, with the role of automatic promotion likely an will be needed to adequately capture the longer-term especially important one. Government policies may have impact of the pandemic on enrollment, attendance, also been fairly effective in protecting attachment to and dropouts. Enrollment rates were likely kept up by school of lower-income youth. several policies. When data become available, it will be important to look at re-enrollment rates to the new The overall small but consistent increase in the out of 2022 academic year. While attendance and enrollment school population rate for 6 to 14 years old in selected rates may have been relatively protected in 2020 and countries may nonetheless be a worrisome signal. The in- 2021, especially for the older cohort, the relaxation of creased numbers of 6 to 14 years old out of school in coun- some policies, the relatively slow economic recovery, tries like Argentina, Colombia, and Brazil, although small, improvement in labor market outcomes and, especially, may point to limited continuous participation and engage- learning losses related to limited attendance and learn- ment in remote schooling, limited priority in school reopen- ing during remote schooling may lead to shocks in 2022 ing, and lack of specific policies to address the attachment to and beyond which could reverse the overall trends in school of this age-group. And these numbers do not consider OOSP population, including and above all for the most numbers of pre-school students which it will be important to vulnerable. 60 Two Years After | Saving a Generation PHOTO BY: © CFALVAREZ / SHUTTERSTOCK.COM Chapter 3 The impact of the COVID-19 pandemic on learning outcomes and skills development The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 61 T he pandemic is expected to have a large impact on learning outcomes and skills development. While the effects on enrollment and to some extent attendance goals set by Sustainable Development Goal (SDG) 4 of the 2030 Agenda for Sustainable Development75. may be not completely clear at this point, the decrease in Results from ERCE 201976 show that, on average, rough- attendance and engagement rates during the pandemic ly one in two 3rd-graders, and three in four 6th-graders are likely to have had important negative implications for do not reach the minimum proficiency level (MPL)77 in learning. And, in turn, learning gaps are likely to have an basic foundational skills. In fact, when we analyze read- impact on future dropouts. This chapter analyzes the im- ing, 44 percent of students in 3rd grade (Figure 3.1a) and pact of the pandemic on learning, considering both simu- 69 percent of students in 6th grade (Figure 3.1c) are be- lated and actual data. Because of the broad-based impact low MPL. In the case of math, figures are even worse: 48 of the pandemic, it also looks at skills development more percent for 3rd grade (Figure 3.1b) and 83 percent for 6th broadly with focus on transferable and digital skills, while graders (Figure 3.1d). These results are appalling, and also documenting the initial impact of the pandemic on they bear two key conclusions. First, they mean that not psychosocial health and well-being, dimensions which af- only the bulk of LAC kids are not learning basic founda- fect schooling, learning and skills acquisition. tional skills by mid-primary, but that the situation only gets worse by the end of that cycle. Second, they highlight that, although numeracy and literacy fare equally bad by mid-primary, the deterioration process by the end of the 3.1 The impact on learning cycle happens faster in numeracy skills. 3.1.1 The state of learning achievement in ERCE 2019 also confirms the wide inequalities in learn- LAC prior to the pandemic: A grim picture ing across the region. For example, the average Costa Rican 3rd grader is 3 times more likely to be above MPL Before the COVID-19 pandemic hit the region, LAC was in reading than his/her Dominican counterpart, or the av- suffering from a deep learning crisis. The vast major- erage Peruvian 3rd grader is twice less likely to fall below ity of students attending compulsory levels of education MPL in math when compared with his/her Paraguayan in the region did not have minimum proficiency levels in peer. By the time kids get to 6th grade, disparities do not go foundational skills, like literacy and numeracy. Evidence away. Countries like the Dominican Republic (98 percent about these worrisome trends got re-confirmed a few below MPL for math) and Nicaragua (87 percent below months back, when UNESCO’s Latin American Laborato- MPL for reading) perform poorly while Peru (61 percent ry for Assessment of the Quality of Education (LLECE, in below MPL for math) and Costa Rica (with 46 percent be- its Spanish acronym) released the latest results from their low MPL for reading) show the best performance. Despite Regional Comparative and Explanatory Studies (ERCE the high variance across countries, data seem to indicate 2019), a large-scale assessment program that involves that school acts as an “equalizer” over time, insofar as the most of the Spanish-speaking countries LAC in addition to size of disparities – as measured by the distance between Brazil74. The study, which has been measuring students’ the best-performing and worst-performing countries in achievement since 1996, underscored the extent of this terms of share of students below MPL - tend to decrease crisis, placing countries far from fulfilling the targets and from 3rd to 6th grade. 74 ERCE measures standardized learning outcomes of students in 3rd and 6th grades covering three main disciplines — reading, mathematics, and natural scienc- es. ERCE assessments are graded around four proficiency levels, with Level I being the lowest and IV the highest. For interpretation and comparability purposes, proficiency level II in ERCE in 3rd grade can be equated to the international Minimum Proficiency Level (MPL), while for grade 6, the MPL is set at Level III. See UN- ESCO-OREALC (2021b). 75 The SDGs were established in 2015 by several United Nations agencies and multilateral/bilateral institutions as a collective of 17 objectives meant to become a springboard for achieving a sustainable future for the world. They have specific targets to be attained by 2030. SDG4, known as the “Education SDG”, aims at “en- suring inclusive and equitable quality education and promote lifelong learning opportunities for all”. See https://sdgs.un.org/goals. 76 UNESCO-OREALC (2021b). 77 MPLs are determined for each subject and each grade differently. For example, in the case of math for 6th grade, being at the MPL means that the student is capable, for example, of solving problems that involve measurement, computation or estimation of areas, or the interpretation of information. MPL in reading for 6th grade means that the student can establish relationships and make inferences from the text. See UNESCO-OREALC (2021b), pp. 14, 16-17. The impact of the COVID-19 pandemic on learning outcomes and skills development 62 Two Years After | Saving a Generation Figure 3.1a. Share of students in grade 3 below MPL Figure 3.1b. Share of students in grade 3 below MPL in reading (%), ERCE 2019 in math (%), ERCE 2019 100 100 80 80 80 73 68 64 65 65 61 59 62 60 60 53 52 50 49 48 46 44 44 47 42 44 43 40 37 36 36 40 37 35 33 30 31 29 28 25 25 24 20 20 0 0 Dom. Rep. Nicaragua Guatemala Panama Honduras Paraguay Argentina LAC El Salvador Ecuador Mexico Colombia Uruguay Cuba Brazil Costa Rica Peru Dom. Rep. Panama Nicaragua Guatemala Paraguay El Salvador Argentina LAC Honduras Colombia Ecuador Uruguay Mexico Costa Rica Brazil Peru Cuba Figure 3.1c. Share of students in grade 6 below MPL Figure 3.1d. Share of students in grade 6 below MPL in reading (%), ERCE 2019 in math (%), ERCE 2019 100 100 98 97 97 94 93 93 87 89 87 84 84 84 83 81 83 83 79 79 77 80 80 74 71 71 69 68 63 62 62 61 60 58 57 56 55 60 51 46 40 40 20 20 0 0 Nicaragua Guatemala Honduras Dom. Rep. Panama Paraguay Ecuador El Salvador LAC Argentina Colombia Mexico Brazil Uruguay Cuba Peru Costa Rica Dom. Rep. Nicaragua Panama Paraguay Guatemala El Salvador Honduras Argentina Colombia LAC Cuba Costa Rica Ecuador Brazil Mexico Uruguay Peru Source: Own elaboration on the basis of UNESCO-OREALC (2021b). MPL is the proportion of students who fall below Level 2 of the assessment in the case of 3rd grade. The ERCE 2019 report also confirms another disappointing gains in learning outcomes between 2013 and 2019. Indeed, fact: there has been limited progress in learning outcomes out of 14 countries that participated in both assessments, in the region. In fact, when compared to TERCE 2013, ERCE only 3 - Brazil, the Dominican Republic, and Peru - were able 2019’s immediate predecessor78,79, results show that, on av- to improve their learning outcomes for both types of students erage, the bulk of countries participating in ERCE showed no (3rd and 6th grades) in both key subjects tested (reading and 78 UNESCO’s LLECE has undertaken four international assessments of this kind to date. The first such assessment, called PERCE, took place in 1995, and had 13 LAC countries participate, all of them being Spanish-speaking one. The second assessment, SERCE, took place 11 years later, in 2006, and increased the number of participating countries to 15, but also adding a separate measurement for the Mexican State of Nuevo León, on top of the one for Mexico. Again here, all countries were of Spanish-speaking origin. The third assessment, TERCE, shortened the period between assessments to 7 years (2013) and again counted with the partici- pation of 15 countries and the State of Nuevo León, but this time around incorporated, for the first time, a non-Spanish-speaking country, Brazil. Finally, the fourth assessment, ERCE 2019, again shortened time between assessments (to 6 years) and added more countries (having, for the first time, a total of 16 countries, 15 Spanish-speaking ones and Brazil). This time around, no sub-national assessment – like the one for Nuevo León in SERCE or TERCE – took place. For details on the four assessments, see https://es.unesco.org/fieldoffice/santiago/projects/llece. 79 To access more data visit www.lleceunesco.org. The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 63 Figure 3.2a. Learning gains in reading and math for students in grade 3, TERCE 2013 and ERCE 2019 30 Peru Dom. Rep. 20 Score di erence in mathematics 2019 vs. 2013 Brazil Paraguay 10 Improvement in math Nicaragua and worsening Improvement in in reading reading and math 0 -10 Panama -20 Worsening in Improvement in reading reading and math and worsening in math Argentina -30 -40 -30 -20 -10 0 10 20 30 40 Score di erence in reading 2019 vs. 2013 Source: Own elaboration based on UNESCO-OREALC (2021b). Countries appearing in the graphs are those that presented statistically significant differences be- tween both assessments for both reading and math (at the 5 percent level). Figure 3.2b. Learning gains in reading and math for students in grade 6, TERCE 2013 and ERCE 2019 50 40 Peru Score di erence in mathematics 2019 vs. 2013 30 Nicaragua Brazil 20 Improvement in math and Dom. Rep. Improvement in worsening in reading reading and math 10 0 -10 Mexico Guatemala -20 Worsening in reading Improvement in reading -30 and math and worsening in math Argentina -40 -40 -30 -20 -10 0 10 20 30 40 50 Score di erence in reading 2019 vs. 2013 Source: Own elaboration based on UNESCO-OREALC (2021b). Countries appearing in the graphs are those that presented statistically significant differences be- tween both assessments for both reading and math (at the 5 percent level). math). Six countries – Colombia, Costa Rica, Ecuador, Hon- Nicaragua, for 3rd-graders, or Paraguay, for 6th-graders – or duras, Panama, and Uruguay – saw no significant progress in also worsening performance, but not in all subjects or grades any of the grades or key subjects. And, most concerning, one – e.g. Nicaragua, with worse results in math for both grades, country – Argentina – showed worse performance in both and Panama (3rd grade only), and Guatemala and Mexico (6th subjects and both years. The remaining countries showed grade only), with worse results in both subjects for specific mixed results, but often for just one subject and grade – e.g. grades (Figures 3.2a and 3.2b). The impact of the COVID-19 pandemic on learning outcomes and skills development 64 Two Years After | Saving a Generation Figure 3.3a. Student achievement differences by gender in Reading, 3rd and 6th grade, ERCE 2019 Reading 3rd grade Reading 6th grade Favorable to boys Favorable to girls Favorable to boys Favorable to girls Argentina 18 Argentina 15 Brazil 12 Brazil 17 Colombia 23 Colombia 13 Costa Rica 12 Costa Rica 11 Cuba 23 Cuba 32 Dominican Rep. 29 Dominican Rep. 36 Ecuador 11 Ecuador 10 El Salvador 21 El Salvador 17 Guatemala 15 Guatemala -1 Honduras 5 Honduras 9 Mexico 11 Mexico 17 Nicaragua 11 Nicaragua 8 Panama 12 Panama 12 Paraguay 21 Paraguay 19 Peru 5 Peru 21 Uruguay 27 Uruguay 17 -30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30 40 Statistically signi cant di erence Statistically signi cant di erence Figure 3.3b. Student achievement differences by gender in Math, 3rd and 6th grade, ERCE 2019 Math 3rd grade Math 6th grade Favorable to boys Favorable to girls Favorable to boys Favorable to girls Argentina -4 Argentina -8 Brazil -7 Brazil -12 Colombia -9 Colombia -6 Costa Rica -13 Costa Rica -9 Cuba 0 Cuba 1 Dominican Rep. 19 Dominican Rep. 6 Ecuador -2 Ecuador 1 El Salvador -4 El Salvador -4 Guatemala -3 Guatemala -14 Honduras -4 Honduras -9 Mexico -6 Mexico 3 Nicaragua -12 Nicaragua -14 Panama -4 Panama -3 Paraguay 2 Paraguay -3 Peru -17 Peru 1 Uruguay 3 Uruguay -4 -30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30 Statistically signi cant di erence Statistically signi cant di erence Source: Own elaboration based on UNESCO-OREALC (2021b). ERCE 2019 results show interesting and fairly con- systematically better in reading than boys, both in 3rd and sistent patterns in connection to gender differenc- 6th grade80 (Figure 3.3a). Second, the clear gender bias es across disciplines tested. Two stylized facts emerge in favor of girls in reading does not replicate at all in the from the analysis. First, on average, LAC girls perform case of math, where there is no discernible gender bias in 80 Actually, the bulk of participating countries show statistically significant differences in favor of girls - 13 of 16 countries in 3rd grade, 14 of 16 in 6th grade. The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 65 Figure 3.4. Factors most strongly associated with student performance in 6th grade, ERCE 2019 Socioeconomic Attendance to Grade Days of study per Parental participation level preschool education repetition week in learning Effect (+) (+) (-) (+) (+) # of countries with statistically 16 16 15 16 14 significant results (reading) # of countries with statistically 16 12 15 15 14 significant results (mathematics) Source: Own elaboration on the basis of UNESCO-OREALC (2021b), Annex 3, p.53. The table only shows those variables: (i) for the student- and family-based asso- ciated factors; and (ii) for which the number of countries with statistically significant effects was at least 12. performance in favor of either boys or girls (Figure 3.3b). In addition, disparities in learning outcomes are also large- Data show that, on average, mixed results are obtained ly explained by differences within classrooms. These large for math performance, where a handful of countries show differences in learning within a classroom bring up another gender differences for specific grades81. enormous challenge for education systems, suggesting that a personalized/targeted approach to learning that adapts to The ERCE 2019 study also reveals that the most sig- each student need is increasingly fundamental for narrow- nificant force driving disparities in learning outcomes ing learning gaps. This personalized approach was the one across and within countries in LAC is the SES of the approach greatly challenged by the massive school closures household. Indeed, by the end of primary education triggered by the pandemic across all countries in the region, (6th grade for the majority of the countries), SES is the but also one being given renewed impetus with the learning only variable that was statistically significant for both recovery agenda, as will be made clear in chapter 5. math and reading for all 16 participating countries (Fig- ure 3.4). From an equity perspective, the results show the 3.1.2 The COVID-19 pandemic persistence of socioeconomic learning gaps and confirm shock and its impact on learning: that differences in learning outcomes between schools A crisis within a crisis are largely a product of socioeconomic segregation within countries. Several other factors also show strong impor- The COVID-19 pandemic generated an “education up- tance, like the number of days of study per week, whether heaval”, disrupting the learning process and generating the child has repeated a grade and attended pre-primary potential learning losses82. This report defines learning education, and the degree of parental participation in the losses as any loss of knowledge or skills and/or deceler- learning process. ation of or interruption to academic progress, most com- monly due to extended gaps of discontinuities in a student’s Last but not least, ERCE 2019 results also show that education. Two main types of learning losses are often dis- between 40% and 50% of learning variability occurs cussed in the literature: “forgotten learning”, which refers between schools and between 50% and 60% with- to the loss of previously acquired learning, and “forgone in schools. The high variability between schools finding learning”, which makes reference to expected learning implies a tight link between quality learning and school that does not take place as schools are closed to in-person characteristics, which vary considerably within countries. learning. To estimate the “forgone” learning, some authors Such diversity within countries represents an enormous estimate a counterfactual using empirical data, such as challenge to educational systems in Latin America, which learning levels attained by previous cohorts (learning loss- need to ensure equal learning opportunities for all stu- es), while others use normative criteria such as grade-level dents despite their socio-cultural and economic contexts. expectations (learning gaps). Even after students return to 81 The situation is fairly even, with 3 of the 16 countries (Costa Rica, Mexico, and Peru) in 3rd grade and 4 of the 16 countries (Argentina, Brazil, Costa Rica, and El Salvador) in 6th grade showing statistically significant differences in favor of boys, whereas only one country in 3rd grade (Paraguay) and a couple of countries in 6th grade (Mexico and Peru), showing that girls outperform boys. 82 WBG (2022c). The impact of the COVID-19 pandemic on learning outcomes and skills development 66 Two Years After | Saving a Generation Figure 3.5. Learning trajectories pre- and post-COVID-19, and implications of current learning losses on future learning Accelerated Learning Recovery Trajectory post-COVID Length of school closure Learning Trajectory pre-COVID Learning Progression Forgetting and Forgone Learning Trajectory learning post-COVID Time Schools Schools Future closing reopening Source: WBG et al. (2021). school, additional learning losses could accumulate, harm- losses published by the WBG84 relied on older macroeco- ing children’s future learning trajectories.83 nomic data and learning assessments, such as he scores from the Programme for International Student Assess- Figure 3.5 illustrates the elements described above us- ment (PISA) from 2018 or before, to prospectively calcu- ing a hypothetical learning progression (slope) given by late learning losses in terms of learning-adjusted years of the learning trajectory pre-COVID. It shows that COVID-re- schooling (LAYS), associated loss of earnings, mean test lated school closures can yield learning losses, both in terms scores (PISA), and share of students below the minimum of forgetting and forgone learning. Such losses can be mea- proficiency levels in PISA. The estimates considered three sured as schools reopen. Additionally, since learning is pro- different scenarios that varied according to the expected gressive, if it is not recovered, students might be pushed length of school closures early in the pandemic and as- towards a new learning trajectory (post-COVID learning tra- sumptions on the effectiveness of each country’s mitiga- jectory) with a flatter slope, which will result in a level of tion efforts85. Now, using observed information on school learning that would be much lower than would have been closures that include data on partial reopening of school expected if students had remained at the pre-COVID rate of systems at the country level, updated assessments data, learning. This difference is referred to in the figure as future and the latest macroeconomic projections, these simula- learning losses. In this context, to redress learning losses, tions can be run for each country retrospectively. For up- students must be put on an accelerated learning recovery dated global simulations86, school closures information is trajectory. Chapter 5 discusses the learning recovery inter- obtained from UNESCO’s Global Monitoring of School Clo- ventions to catch up on lost learning. sures database87, and the different scenarios are differenti- ated by their assumption on the extent of school closures Simulating the expected in partially opened systems (50%, 25%, and 15% for the learning losses for LAC optimistic, intermediate, and pessimistic scenarios, re- spectively) and the effectiveness of mitigation efforts (high, Updated simulations using observed data on length of medium, and low, for the optimistic, intermediate, and pes- school closures to date show significant learning losses simistic scenarios respectively). For these regional simu- in the region. Previous estimations on simulated learning lations, we further improve these estimates in two ways. 83 WBG et al. (2021). 84 For previous publications in chronological order, see Azevedo (2020), Azevedo et al. (2020), WBG (2021a), and WBG et al. (2021). 85 See Azevedo et al. (2020) for details. 86 See Azevedo et al. (2022) for details. 87 Data source: http://COVID19.uis.unesco.org/global-monitoring-school-closures-COVID19/. The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 67 First, we calculate the length of school closures by Febru- efforts in certain countries to provide remote learning, ary 2022 aggregating data from different sources, including like Plan Ceibal in Uruguay, and to fully open schools to the UNESCO calendar, the UNICEF’s Monthly monitoring in-person classes seem to have paid off, with Uruguay los- of school closures88, and country-specific web research to ing only 0.88 LAYS in the same scenario. In relative terms, get more accurate data89. Second, for simulations on test the loss in learning represents a bigger share in countries scores and levels of proficiency, our simulations incorpo- with lower LAYS prior to the pandemic. For example, Gua- rate the latest data on student learning from the ERCE 2019 temala and El Salvador could lose 16% of LAYS relative to assessments for a subset of countries in the region90. This their baseline in the intermediate scenario, while Hondu- allows us to see the effects on mean test scores and distri- ras could lose up to 20% of LAYS. bution of students within the different levels of proficiency for math and reading among third and sixth graders. Using These learning losses translate into a significant de- these updated data, we simulate (i) the effect on LAYS and crease in earnings and productivity equivalent to a de- earnings91 using the UNESCO calendar scenarios and the crease of about 12% in lifetime earnings for a student at aggregated WBG calendar92, and (ii) effect on ERCE mean school today under an intermediate scenario. The loss scores, levels of proficiency and share of students below of learning can be quantified in terms of lifetime earnings minimum proficiency for the aggregated WBG calendar. using evidence on returns to schooling, life expectancy and labor market variables. In the optimistic scenario, the av- Estimates using different assumptions on the extent of erage student in LAC from the cohort in school today could partial reopening show a loss of between 1 and 1.8 LAYS. lose $931 (in 2017 dollars of purchasing power parity or Using the UNESCO’s calendar, in the optimistic scenario, PPP) average annual earnings, equivalent to $16,993 in with partial reopening assumed to be 50% open, there is lifetime earnings and 7% of projected lifetime earnings. In an average loss of 1 LAYS in the region, from a baseline the intermediate scenario, the average student in LAC from of 7.8 years (see Figure 3.6). In the intermediate scenario, the cohort in school today could lose $1,565 (in 2017 PPP with partial reopening assumed to be 25% opened, losses dollars) average annual earnings, equivalent to $28,556 in reach 1.5 LAYS. Lastly, in the pessimistic scenario with only lifetime earnings and 12% of projected lifetime earnings. 15% opened, losses reach 1.8 LAYS. Using the aggregated This effect compounds the economic crisis already result- WBG calendar, simulations show that LAYS would decrease ing from the pandemic, and countries may suffer significant from 8.2 to 6.9 (assuming high mitigation effectiveness), losses in human capital and productivity. It is expected to 6.5 (assuming medium mitigation effectiveness) and that learning losses in LAC may translate into an aggregate 6.4 (assuming low mitigation effectiveness). These values economic cost of foregone earnings of US$1.5 trillion lost approximate more those of an optimistic assumption on (in 2017 PPP dollars) in the optimistic scenario, equivalent partial school openings in terms of the percentage of the to 16% of the regional GDP, and US$ 2.3 trillion lost and school time schools remained closed. Thus, optimistically, 24.7% of the regional GDP, in the intermediate scenario. children are one learning-adjusted year of schooling be- Using the aggregated WBG calendar data, and assuming in- hind in the region on average, despite attempts to reopen termediate effectiveness of mitigation efforts, the average schools and provide distance learning. student could lose $1,374 (in 2017 PPP dollars) average annual earnings, equivalent to 10% of projected lifetime These averages hide large variations between coun- earnings. tries, with larger relative losses for worse-off countries. As in previous simulations, the loss in absolute terms Simulations reveal that learning outcomes as mea- could be higher for countries that were better off before sured by mean ERCE scores decrease significantly for COVID-19. Looking at the scenarios from the UNESCO cal- both math and reading and 3rd and 6th grades. For the endar, in Costa Rica, Ecuador, and Mexico, LAYS could drop 16 participating countries in the region, baseline (2019) by almost 2 years in the intermediate scenario. However, ERCE scores were on average 718 and 715 for 3rd grade 88 Data source: https://www.unicef.org/lac/respuesta-educativa-de-alc-frente-al-COVID-19/. 89 This aggregated WBG calendar divides the “Partially Closed” classification further into three: mostly opened, partially closed/opened, and mostly closed. Each of these classifications is converted into a measure of the percentage of time a school system was closed: 75% for mostly closed, 50% for partially closed, and 25% for mostly open. 90 The 16 countries included are: Argentina, Brazil, Colombia, Costa Rica, Cuba, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nica- ragua, Panama, Paraguay, Peru, and Uruguay. 91 LAYS and associated earnings losses simulations still use the original simulation model based on PISA 2018 or older data. 92 This calendar includes information on 3 more countries: St. Kitts & Nevis, St. Lucia, and St. Vincent & the Grenadines. The impact of the COVID-19 pandemic on learning outcomes and skills development 68 Two Years After | Saving a Generation Figure 3.6. Simulated loss in learning-adjusted years of schooling (LAYS) due to COVID-19 in LAC 10 9.4 8.8 9.0 9.1 9 8.4 8.4 8.6 8.6 8.6 8.7 8.0 8.3 8 7.6 7.8 7.9 6.7 6.8 7.0 7.1 7 6.3 6.5 6.6 7.7 7.9 6.1 6.2 7.4 6 7.0 7.0 6.9 6.7 6.8 7.1 6.5 6.6 6.7 6.8 5 5.9 6.3 6.1 5.6 5.4 5.5 5.6 4 4.6 5.2 4.7 4.8 3 2 1 0 Honduras Haiti Guatemala Panama Dominican Republic Nicaragua Guyana Paraguay Jamaica El Salvador LAC Region Brazil Dominica Grenada Argentina Antigua and Barbuda Uruguay Colombia Peru Ecuador Mexico Costa Rica Trinidad and Tobago Chile Baseline Optimistic Intermediate Pessimistic Source: World Bank estimates based on Azevedo et al. (2022). Note: The parameters used here match global simulations based on the country’s income level group. Simulations presented are based on UNESCO’s calendar until February 2022. Optimistic, Intermediate, and Pessimistic scenarios vary in their assumption of actual school closures during partial reopening (50, 25, and 15 % closed, respectively) and effectiveness of mitigation strategies (high, medium, and low, respectively). math and reading, respectively, and 725 and 715 for 6th Figure 3.7. Simulated learning loss effects on mean grade math and reading, respectively93. The mean ERCE ERCE 2019 scores, by grade and subject in LAC (%) scores are expected to decrease by around 6.3% (or 45 800 points) in both grades and subjects, given our best esti- 725 718 715 715 mate on the actual length of school closures from the ag- 673 670 680 669 700 gregated WBG calendar and a medium level of mitigation effectiveness for distance learning. Figure 3.7 presents 600 the decrease in the average ERCE scores corresponding to these 16 countries by grade and subject. To put things in 500 perspective, the simulation results for 3rd grade imply that 400 every country would have worse mean scores than in 2013 and results would fall to levels of more than 10 years ago, 300 considering average TERCE 2013 scores of respectively 700 for reading and 709 for math. 200 Beyond the mean, the distribution of students along 100 ERCE scores, that is the percentage of students belong- ing to each of the 4 categories will also be greatly af- 0 Math Reading Math Reading fected by the pandemic. Figure 3.8 shows the percentage Third Grade Sixth Grade of students in each of the 4 proficiency levels of ERCE by Baseline Intermediate grade and subject at baseline (based on ERCE 2019 scores) and simulated using school closures from the aggregate Source: World Bank estimates based on Azevedo et al. 2022. Note: The parameters used here match global simulations based on the coun- WBG calendar and assuming an intermediate effective- try’s income level group and assume an intermediate effectiveness of miti- ness of remote learning. For all tests, the share of students gation strategies. The length of school closures is defined on a country level based on an aggregation of UNESCO, UNICEF, and other data sources until in level 1 is expected to increase the most. The share of February 2022. students in proficiency level 1 increases by 17 percentage 93 The country-level mean scores between official UNESCO results and the baseline values for ERCE 2019 used in the simulation tool differ slightly because the simulation tool estimates baseline values for 2019 using grouped data. Scores are population weighted averages. The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 69 Figure 3.8. Simulated learning loss effects on the ages and in reading. The minimum proficiency level is distribution of ERCE 2019 proficiency levels, by defined differently in each grade.94 Figure 3.9 shows the grade and subject in LAC (%) resulting share of students below MPL at baseline and after the pandemic for each grade and subject. In most coun- 100% 6% tries, the share of students below MPL is expected to in- 7% 8% 7% 11% 14% crease more for students in third grade (panel A and B) 90% 21% 10% 21% 12% and in reading (panels B and D), which is consistent with 15% 18% 80% expectations on the effectiveness of distance learning for 26% 18% younger students and in reading. In the region95, on aver- 70% 30% 17% 21% 25% age, the share of students below MPL for reading in sixth 60% 18% grades increases from 62% to 82% (20 p.p.) and from 74% 36% 50% to 84% (10 p.p.) in math. For third graders, the share of 50% 24% students below MPL increases from 37% to 50% in reading 17% 40% 44% (13 p.p.) and from 39% to 56% in math (17 p.p.). Recent forthcoming updates of learning poverty estimations do 30% confirm that Latin America is the region in the world with 56% 54% 50% 20% 39% the starkest increase in children unable to read a simple 37% 38% 32% passage with comprehension by age 10 (see Box 3.1). Ex- 10% 18% pected losses are already showing to be a reality according to the emerging evidence on actual learning losses. 0% Intermediate Intermediate Intermediate Intermediate Baseline Baseline Baseline Baseline Actual learning losses: Reviewing the evidence to date Math Reading Math Reading 3er grade 3th grade 6th grade 6th grade Learning assessments are a vital tool to provide a clear Treshold I Treshold II Treshold III Treshold IV picture of students' current learning levels. They are es- sential to attempt to measure the effect of the pandemic on Source: World Bank estimates based on Azevedo et al. (2022). Note: The parameters used here match global simulations based on the coun- learning losses, and to inform the formulation of adequate try’s income level group and assume an intermediate effectiveness of miti- and evidence-based targeted and remedial programs. As- gation strategies. The length of school closures is defined on a country level based on an aggregation of UNESCO, UNICEF, and other data sources until sessments also allow periodical tracking of the students to February 2022. gauge their progress and learning recovery after the crisis. During the year 2020, most countries in the region de- points in math for third graders (from 39% to 56%) and cided to postpone or suspend their national learning 13 percentage points in reading (from 37% to 50%). For evaluations due to the pandemic. During 2020 there sixth graders, the share of students in level 1 increases by was a tendency to maintain certification and/or student 16 percentage points in math (from 38% to 54%) and by selection exams, and to suspend or postpone national 14 percentage points (from 18% to 32%) in reading. The evaluations for monitoring purposes. As shown in Chap- percentage of students in high-scoring levels 3 and 4 de- ter 5, many countries canceled the scheduled national creases in all grades and subjects. The higher percentage evaluations (Argentina, Belize, Dominican Republic, Costa of students in level 1 poses an especially strong challenge Rica, Guatemala, Honduras, Panama, and Peru) or evalu- to teachers as they must cater to the need of this increas- ated fewer grades or subjects. Brazil, Uruguay, and El Sal- ing proportion of students in the classroom. vador continued with the national evaluation as planned in 2020. However, in 2021 most countries resumed the This changing distribution of scores has consequences measurement of learning outcomes. The challenges and on the share of students below MPL, which is expect- innovations adopted by LAC countries for different types ed to dramatically increase, especially for the younger of assessments will be discussed in Chapter 5. 94 As explained above, for third graders, a student is defined as being below minimum proficiency if he or she is in level 1. For sixth graders, a student is defined as being below minimum proficiency if he or she is in level 1 or level 2. 95 For this exercise, we calculated a population weighted average that considers the number of students in each of the included countries. The impact of the COVID-19 pandemic on learning outcomes and skills development 70 Two Years After | Saving a Generation Box 3.1. Learning poverty in LAC: A serious issue, deepening with COVID-19 The learning poverty indicator measures the share of children who are unable to read and understand a sim- ple text by age 10. High rates of learning poverty are an early signal that education systems are failing to ensure that children develop critical foundational skills. This makes it much harder for children to acquire the technical and higher-order skills needed to thrive in increasingly demanding labor markets and for countries to develop the human capital needed for sustained economic growth96. The high rate of learning poverty was a global rising Figure B.3.1.1. Learning poverty, globally and by problem before the pandemic and expects to deepen region (%), 2015 – 2022 (est.)* now after COVID-19. The average global learning pov- 100.0 erty rate was estimated to affect 57 percent of 10-year- 90.0 olds in low- and middle-income countries (LMICs) in 80.0 79.0 2019, up from a 53 percent value in 2015. The latest 70.0 70.0 global simulations, just released by the World Bank 60.0 within the scope of a joint effort by many international 52.6 50.0 50.8 52.3 agencies (WBG 2022d), project a 23 percent increase, 40.0 reaching a dismal 70 percent. In other words, after 30.0 two years of living under the effects of the pandemic, 20.0 learning poverty in low- and middle-income countries 10.0 would be expected to affect 7 out of 10 10-year-olds. A 0.0 sobering trend. 2015 2019* 2022* East Asia and the Paci c Once again here, the pandemic would be dispropor- Middle East and North Africa tionately affecting the LAC region. Learning poverty Global (LMICs) was estimated to affect roughly half of all children in Europe and Central Asia the region. In line with global trends, it was also pre- South Asia senting a negative trend since 2015 - increasing from Latin America and the Caribbean an average 51 percent in 2015 to 52 percent in 2019. Sub-Saharan Africa Prior to the pandemic, nonetheless, LAC showed av- *Note: Numbers for 2022 are based on simulations under a pessimistic sce- erage learning poverty rates which were only higher nario. The global figure is for all low- and middle-income countries. Regio- than two regions - EAP (35 percent) and ECA (10 per- nal and global figures are all population-weighted averages. For the EAP re- gion, the 2015 and 2019 averages are not directly comparable, due to major cent). According to the new simulations, the deepening of changes in the country composition and assessments used for the two years. learning poverty in LAC would be so dramatic that the re- Sources: Azevedo et al. (2022), WBG (forthcoming b). gion would now experience the largest relative increase in learning poverty (26.7 percentage points) in the world. In fact, simulations forecast that learning poverty in LAC would reach 79 percent by 202297, surpassing the average for low- and middle-income countries and overtaking regions like MENA (Figure B.3.1.1). Increases in learning pov- erty would be largest in those regions – SA and LAC - where school closure spells were the longest. In these regions, the share of children in learning poverty is expected to have increased by more than half in just the past two years. Sources: Azevedo et al. (2022); WBG (forthcoming b). 96 WBG (2022d). 97 It is important to highlight that the calculation of learning poverty is based on the SERCE scale (which defined the minimum proficiency level as those students reaching Level 3 in language (or a score above 514 points)), so these estimations, while consistent in the magnitude of the increase, are not directly comparable with the simulations presented in Figures 3.7 through 3.9, which use the TERCE scale for comparability with 2013. The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 71 Figure 3.9. Simulated learning loss effects on the share of students below MPL in ERCE 2019, by grade and subject in LAC (%) Panel A: Math 3rd grade Panel B: Reading 3rd grade 100% 100% 80% 80% 60% 56% 60% 50% 40% 40% 20% 20% 0% 0% Argentina Brazil Colombia Costa Rica Cuba Dominican Republic Ecuador El Salvador Guatemala Honduras Mexico Nicaragua Panama Paraguay Peru Uruguay Group average Argentina Brazil Colombia Costa Rica Cuba Dominican Republic Ecuador El Salvador Guatemala Honduras Mexico Nicaragua Peru Panama Paraguay Uruguay Group average Baseline Intermediate Baseline Intermediate Panel C: Match 6rd grade Panel D: Reading 6rd grade 100% 100% 84% 82% 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% Argentina Brazil Colombia Costa Rica Cuba Dominican Republic Ecuador El Salvador Guatemala Honduras Mexico Nicaragua Panama Paraguay Peru Uruguay Group average Argentina Brazil Colombia Costa Rica Cuba Dominican Republic Ecuador El Salvador Guatemala Honduras Mexico Nicaragua Panama Paraguay Peru Uruguay Group average Baseline Intermediate Baseline Intermediate Source: World Bank estimates based on Azevedo et al. (2022). Note: The parameters used here match global simulations based on the country’s income level group and assume an intermediate effectiveness of mitigation strate- gies. The length of school closures is defined on a country level based on an aggregation of UNESCO, UNICEF, and other data sources until February 2022. Regional and international assessments were mostly comparable evaluations over time is still a challenge in kept on track. The high-stakes assessments of the Carib- several LAC countries. Additionally, the results of most bean region, developed by the Caribbean Examinations of the 2021 evaluations are still pending publication. This Council (CXC), were applied in 2020 and 2021. The PISA has resulted in having few data and studies at this stage global assessment was postponed, and its next application that can help measuring the learning loss generated by the is planned for 2022. However, PIRLS (Progress in Interna- two years of the pandemic. tional Reading Literacy Study) was applied in-person in 2021. National and regional assessments undertaken in 2020 (and 2021 when available) can help provide a first snap- While it is good news that many assessments took place shot of trends in learning, with caveats. Colombia imple- in 2021, evaluation challenges persist and carrying out mented SABER examinations in 2020 and 2021, allowing The impact of the COVID-19 pandemic on learning outcomes and skills development 72 Two Years After | Saving a Generation Figure 3.10. Colombia: Share of Grade 3 students at Figure 3.11. Colombia: Share of Grade 3 students each reading level by frequency of attendance (%), at each math level by frequency of attendance (%), 2020 2020 100% 100% 4.5 7.1 8.4 5.9 6.9 12.7 90% 16.7 90% 7.4 12.3 13.8 80% 80% 14.6 14.5 15.9 16.4 70% 70% 24.8 21.6 60% 60% 24.6 25 50% 50% 40% 30.5 30.3 40% 67.4 67.9 30% 24.8 30% 47.6 20% 20% 25.4 26 10% 10% 17.2 6.7 2.9 1.6 0% 0% Very low Low O en Very low Low O en Frequency of class attendance Frequency of class attendance Level 1 Level 2 Level 3 Level 4 Level 5 Level 1 Level 2 Level 3 Level 4 Level 5 Source: INEEd, 2021. Source: INEEd, 2021. for some comparison, although with caveats98, between significant reduction in learning outcomes, expressed pre- and post-pandemic outcomes (see Box 3.2). Learning as average test scores. Better overall results in Uruguay scores slightly decreased, continuing a long-term trend, were partly related to higher readiness for remote learn- with persistent gender, socioeconomic, private-public, ing through Plan Ceibal, which broadly increased Inter- rural-urban disparities. The magnitude of the decrease is net and computer access for students and teachers.100 however difficult to interpret in this context given that this The relevance of Plan Ceibal gained greater visibility in is a high-stake exam which was subject to important cohort the context of the pandemic due to the increase in the effects in 2020 and 202199. Results on the high-stake Carib- use of platforms. Almost 90 percent of teachers indicat- bean Secondary Education Certificate (CSEC) examinations ed that they were linked to students in the context of the in Jamaica show that the share of students with a passing pandemic through Plan Ceibal platforms.101 The prompt- grade in Mathematics dropped from 47 percent in 2018 to er reopening of schools also contributed to better re- 38.2 percent in 2021. sults. Students from 3rd grade who attended school more frequently, showed higher levels in reading. Of the chil- Outcomes from Uruguay illustrate the importance of dren with low attendance, only 19.4% achieved reading readiness for remote learning and school reopening. levels 4 and 5, compared to 33.1% of children who usu- A comparison of the 2017 and 2020 low-stakes Aristas ally/often attend class (Figure 3.10). A similar result is assessment in Uruguay (the test uses a representative observed for mathematics in 3rd grade: 11.4% of children sample of all 3rd and 6th grade students) shows no with very low attendance were in mathematics levels 4 98 It is worth noting that, while the SABER 11 test is comparable in its applications in 2019, 2020 and 2021 (pre- and post-pandemic), the SABER 3, 5 and 9 tests were applied only through 2017, for which there are pre-pandemic estimates, but not post-pandemic. The latter results are therefore comparable only through 2017 because the test was significantly changed in 2018. In fact, in 2018 the SABER 3, 5, 7, and 9 tests were re-designed to be standardized and comparable over time. In 2019, a pilot application for all four SABER tests was carried out. The full application of the 4 revised tests was first carried out in 2021. As a result, comparability between the pre-pandemic and post-pandemic applications of SABER 3, 5, 7, and 9 tests is necessarily limited. 99 In Colombia, official data from the Colombian Institute for the Education Quality Assessment (ICFES), which is in charge of the national evaluation tests, shows that in 2020 the number of students that participated in the national evaluation for eleventh grade was 2% lower than in 2019. Nonetheless, in 2021 the participa- tion rate in the evaluation was even higher than in 2019 (4% higher) in part given by the fact that people who did not do the test in 2020, did it in 2021, which is suggestive of temporary rather than permanent effects on schooling. 100 100 percent of primary and lower secondary schools (3,023 schools) had Wi-Fi connection and access to digital devices (laptops and tablets). See https:// www.ceibal.edu.uy/es/articulo/ceibal-en-cifras. 101 Carbajal et al. (2022). The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 73 Box 3.2. Assessing learning in difficult times of the pandemic: Colombia’s culture of evaluation Colombia demonstrated commitment to an evaluation culture by applying the SABER test in the context of the pandemic to give continuity to the measurement of its students’ competencies. Due to the pandemic and contin- ued school closures, many countries postponed or canceled the learning assessments that were scheduled. However, Colombia maintained the application of the SABER exam during 2020. The system for evaluating the results of the quality of education in Colombia is led by the Colombian Institute for Education Quality Assessment (ICFES). Be- cause the SABER exams have national coverage, conducting the exam in 2020 involved a major logistical challenge. The places where the face-to-face exams were applied were, in most cases, schools that had been closed for most of the year and had to be reopened in November solely for the application of the exam and complying with the biosafety protocols. By law, state exams are applied twice a year to evaluate the education given to those who finish upper sec- ondary education, called SABER 11 (Grade 11). For SABER 11, the application of the first semester (which takes place approximately in March) mainly evaluates secondary education institutions whose school calendar begins in August and ends in June of the following year (calendar B), while the application of the second semester (which takes place between September and October) mainly assesses students from institutions whose school calendar begins in January and ends in November (calendar A). The population enrolled in calendar B is a minority compared to that of calendar A and is also largely made up of private and bilingual institutions. 100% of the country’s public institutions belong to calendar A. Temporary adaptations were made to the SABER 11 test to achieve its application in 2020, without compromis- ing the quality, comparability, or rigor of the exam. Generally, the exam is carried out in two sessions of four hours and thirty minutes each. In 2020, the tests were applied in a single session and with a smaller number of questions, for which it was necessary to reduce the number of questions from 278 to 188, with a total application time of 5 hours and 30 minutes. This change was rigorously evaluated to ensure that comparability with results from other applica- tions was not affected. In 2021, the ICFES resumed the usual application of the exam. Figure B.3.2.1. Colombia: Average SABER-11 test Figure B.3.2.2. Colombia: Average SABER-11 test results, 2015-2021 results by SES, 2019-2021 340 350 326 321 318 321 320 316 315 300 310 300 250 Average Score Average Score 280 200 264 262 260 256 258 253 252 150 240 250 100 220 50 200 2015 2016 2017 2018 2019 2020 2021 0 Year 2019 2020 2021 Calendar A Calendar B SES 1 SES 2 SES 3 SES 4 Source: Own elaboration with data from ICFES, using SABER-11 data (2021). Note: SABER 11 evaluated 486,271 Grade 11 students and 55,207 “Cycle 6” students. Cycle 6 corresponds to the last year of secondary education for adults. Since the focus of this section is on children and adolescents, results from Cycle 6 were not included, only those from Grade 11. ICFES applies SABER tests to assess students learning in elementary school (grade 3 and grade 5), lower secondary (grade 7 and grade 9) and upper secondary (grade 11). Only SA- BER-11 results are comparable pre- and post-pandemic. (Box continues on next page) The impact of the COVID-19 pandemic on learning outcomes and skills development 74 Two Years After | Saving a Generation Box 3.2. Assessing learning in difficult times of pandemic: Colombia’s culture of evaluation (continued) Regarding the results, in 2020 and 2021 there is a slight decrease in the average score with respect to the results of 2019. The maximum score of the SABER 11 exam is 500, where the scores are weighted in each of the 5 competencies that the exam examines: mathematics, critical reading, social and citizenship, natural sciences and English. In 2020 the average Calendar A score dropped one point (from 253 to 252 points) and in 2021 it dropped an additional two points (score 250)-Figure B.3.2.1-. The drop was strongest for Schedule B, whose average score fell 11 points between 2019 and 2020 (from 321 to 310). However, in 2021 it increased again by 5 points (score of 315). It is important to highlight that these variations are not statistically significant with a confidence level of close to 95%. Last but not least, some gaps in results are observed for some groups. Figure B.3.2.2 illustrates the scores (adding calendar populations A and B) for each socio-economic status between 2019 and 2021. As the socioeconomic level increases, the average global score also increases. In 2021, the students of SES 1 had an average of 222 points, while those of SES 4 obtained an average of 301. The results show a very wide and significant difference (79 points) in the average score between the students of the level highest and lowest socioeconomic status. Source: ICFES, Colombia. SABER 11, 2021. and 5, compared to 27.2% of children who usually/of- results show learning losses equivalent to scores from 10 ten attend class102 (Figure 3.11). The pandemic however years ago or more: 196 (2021) in math and 194 (2021) in interrupted the educational process in an unequal way Portuguese (Figures 3.12 and 3.13). The drop is starker depending on the context: those who were not (came for primary education students. Assuming the pre-COVID back later) in the classrooms were, to a greater extent, learning trajectory (2011-2019) is maintained, recover- children in more vulnerable situations. As a result, there ing from the learning losses would take between four and appears to have been an increase in the academic gap fifteen years104. This study is the starkest illustration so between lower- and higher-income students. These data far that expected learning losses are real and simulations are also nicely illustrative of the expected impact of lim- are not far from reality, with, notably, a decrease of about ited attendance on learning in other countries, knowing 13% in reading in grade 5. And this is especially worri- that the situation is likely to be much worse given the some knowing that these data refer to a wealthy State in higher preparedness of Uruguay for distance learning. Brazil. A second study from São Paulo concluded that un- der remote learning, students had a setback equivalent to Studies from LAC regarding actual learning losses are 75% of in-person learning with a risk of school drop-out still scarce and not all studies have the same statisti- increasing by 365 percent in the wake of the pandemic 105. cal rigor, but the available evidence points to signif- Additional data from a group of Brazilian states indicate icant learning losses. Table 3.1 summarizes the main that the share of second graders off track to becoming flu- findings on learning losses for selected studies from the ent readers (who can only read a maximum of 9 words region. Two robust studies from São Paulo (Brazil) show in one minute) increased from 52 percent in 2019 to 73 pronounced learning losses. The first study uses stan- percent in 2021106. dardized test scores to compare learning before and after school closures.103 The study finds learning losses for all A study from Chile compares child development indica- assessed grades in math and reading. For 5th grade, the tors for pre-primary students, finding children affected 102 INEEd (2021). 103 At the end of 2021 the National Institute of Educational Studies and Research (INEP) started the in-person application of the Basic Education Assessment Sys- tem (SAEB). The application included public school students from 5th and 9th grades of elementary school and 12th and 13th grades of high school. These same grades were also evaluated in a sample format for private schools. The areas evaluated were Portuguese (language) and mathematics. Basic Education Assessment System analyzed a sample from 5th, 9th and 12th grades. 104 São Paulo Education Secretariat (2021). 105 Cossi et al. (2021). 106 Fundação Lemann (2021). The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 75 Figure 3.12. Brazil: National SAEB score results Figure 3.13. Brazil: National SAEB score results in in language: Time trend by grade in São Paulo, mathematics: Time trend by grade in São Paulo, 2011-2021 2011-2021 290 290 268 270 263 270 268 255 250 245 250 250 248 249 230 230 210 210 213 190 190 196 192 194 170 170 150 150 2011 2013 2015 2017 2019 Start 2011 2013 2015 2017 2019 Start 2021 2021 Grade 5 Grade 9 Grade 12 Grade 5 Grade 9 Grade 12 Source: São Paulo Education Secretariat (2021). Note: The grades are graphed together for illustration, although the scales are not comparable across grades. by the pandemic scored worse than the pre-pandemic rural areas) and results, although also still limited, have cohort in three out of four developmental areas (lan- tended to penalize girls. Some studies from the OECD (Or- guage, social-behavioral skills, and general develop- ganisation for Economic Co-operation and Development) ment)107. Other studies for Colombia108 and Mexico109 countries show similar patterns.110 show significant learning losses, especially affecting dis- advantaged groups. In Mexico, a recent study on the State Evidence confirms that younger students and early of Guanajuato supported by the World Bank presents ro- grades have been disproportionally affected. Results bust preliminary evidence of a decrease in learning out- from São Paulo using SAEB, the Basic Education Assess- comes (Box 3.3). ment System, show much starker declines for grade 5 than for grades 9 and 12. Data from Mexico show a steeper drop Growing evidence from LAC and low- and middle-in- in reading results for early ages (10 and 11 years) than come countries confirms learning losses have unequal- for older ages (12 to 15 years). In India, the proportion of ly impacted vulnerable groups. Based on recent studies, children in grade 3 enrolled in public schools who can do Table 3.2 summarizes the effect that certain characteris- at least subtraction (proxy for grade level) dropped from tics have had on learning losses to date. On average, ear- 23.5% in 2018 to 16% in 2020. In contrast, for grade 7, the ly grades, younger children and children from lower SES proportion of children who can do division increased in have been disproportionally affected with greater learn- that same period (36.1% to 36.6%). There are larger loss- ing losses. There is limited evidence by location (urban vs. es in grades 3 and 5 than in grade 7. 107 Abufhele et al. (2021). 108 García et al. (2021). 109 Hevia et al. (2022). 110 In Belgium, for example, a study finds increases in inequality both within and across schools, driven by large learning losses in schools with a high proportion of disadvantaged students (Maldonado and de Witte, 2021). A study from Switzerland showed that the learning pace of primary school students was much lower in remote learning than the learning pace of secondary school students (Tomasik et al., 2021). A study from the United States found larger declines in math, and the earlier grades, and marginalized and economically disadvantaged students had larger declines relative to their peers (Lewis et al., 2021). Recent evidence from the Basque Country (Spain) finds that factors linked to schools account for learning loss differences, observing a large decline in learning for public schools, as well as learning in private schools with prior low performance (Arenas and Gortázar, 2022). The impact of the COVID-19 pandemic on learning outcomes and skills development 76 Two Years After | Saving a Generation Box 3.3. Measuring learning and promoting alliances for learning recovery: The case of Guanajuato The state of Guanajuato (GTO), in Central Mexico, is a great example of an education system that has developed an Education Management Information System (EMIS) and a learning assessment culture for evidence-based policymaking. Against such background, GTO's Ministry of Education (SEG) decided to build on this information-driv- en policy approach to guide critical actions for recovery after the pandemic111. In doing so, they became Mexico’s only state to place a strong emphasis in the use of student learning outcomes (both summative and formative) as a first stage to guide the schooling and learning recovery strategy in coordination with other human development public institutions and in alliance with the private sector112,113. Figure B.3.3.1. Guanajuato: RIMA average scores by level, grade, and subject, 2020 vs. 2021 Primary Level Secondary Level 50 49 0% 50 0% 45 45 44 44 41 41 40 39 40 38 38 38 37 35 -5% 35 -5% - -7.32% -7.89% 25 -10.20% -10% 25 -9.76% -10% -11.11% -11.36% -12.50% -15% -15% -15.56% 0 -20% 0 -20% Spanish Math Spanish Math Spanish Math Spanish Math Fi h Grade Sixth Grade Eigth Grade Ninth Grade 2020 2021 % change 2020 2021 % change Source: Own elaboration based on RIMA data. Note: The scale of these scores goes from 0, for students that answered all items incorrectly, to 100, for students that answered all items correctly. The percent change in scores from baseline is presented in the yellow boxes. Prior to the pandemic, the SEG had developed and piloted a standardized test called Data Collection for Im- proving Learning Outcomes (Recopilación de Información para la Mejora de los Aprendizajes or RIMA). The effort emerged at a time when the national assessment PLANEA for basic education had been interrupted indefinitely. RIMA was piloted in six municipalities114 of GTO one week before schools closed in March 2020. The pilot covered two sub- jects (Spanish and mathematics) and was administered in four grades: 5th and 6th grades of primary education, and 8th and 9th grades of lower secondary education. The test was later applied state-wide (over 600,000 students) when students began returning to in-person classes at school, between November and December 2021. This second assess- ment round of RIMA was administered to grades 4, 5, and 6 in primary education, and grades 7, 8, and 9 in lower secondary education. The test also included a module on socioemotional skills. The unique timing of the application for both tests therefore allows for the measurement of learning losses due to school closures during the pandemic in these six municipalities. Average proportion of correct answers for both rounds by level, grade (only those that participated in both rounds), and subject are shown in Figure B.3.3.1. (Box continues on next page) 111 GTO’s schooling and learning recovery strategy includes 4 main actions: i) re-enrolling and retaining students; ii) improving teaching practices; iii) recovering and accelerating learning; and iv) accompanying parents to promote learning at home. 112 Nuevo León (Northern Mexico) has been assessing students' learning outcomes from 2nd to 12th grade. 113 While results from GTO cannot be generalized nationwide, many of the socio-demographic features of the country (e.g., rural communities, small and big cit- ies, indigenous and non-indigenous groups, people from all income levels, etc.) are, on average, fairly well captured by this state. 114 The six municipalities were Jaral del Progreso, Moroleón, Salamanca, Uriangato, Valle de Santiago, and Yuriria. The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 77 Box 3.3. Measuring learning and promoting alliances for learning recovery: The case of Guanajuato (continued) Preliminary results show a decrease in average scores for all grades and subjects, with the largest learning losses, on average, being observed at the primary level. In fact, learning losses range from 10-16% from baseline (in primary) to 7-11% (in lower secondary). In addition to learning losses, GTO is also experiencing an increase in student drop-out rates. The SEG has estimated that 70,000 students have dropped out since the pandemic started115. The WBG is providing support to GTO to: (i) assess post-pandemic learning levels and measure learning losses; (ii) communicate results to relevant stakeholders; and (iii) use these data to improve decision-making and teacher prac- tices in the State to support the schooling and learning recovery agenda. Final data analysis is expected to take place in April 2022, and results to be disseminated with schools, teachers, students, and parents immediately thereafter. According to SEG, RIMA is just the starting phase of the education recovery strategy. In fact, a Social Pact for Edu- cation is being launched where civil society, the private sector, and religious institutions are joining forces with the public sector to ensure that students return to schools and accelerate learning. Interventions to support the learning recovery were planned to start in May 2022. Source: Own elaboration based on RIMA data. Table 3.1. Selected studies for LAC on learning losses Country Scope Subject Learning Metric Grades/ Results Areas Level/Ages Brazil 1 Sub- Math and SAEB 2019 and Sample G5, G9, G12 Learning losses for all assessed grades in math national: Portuguese Assessment in 2021 and reading. Losses equivalent to between 4-15 São Paulo years of schooling in São Paulo. Brazil 2 Sub- Math and Standardized test scores G6, G7, G8, Students under remote learning had a setback national: Portuguese G9, G10, G11, equivalent to 75% of the in-person learning. São Paulo G12 Chile 3 Sample Child Peabody Picture Vocabulary Pre-primary Children affected by the pandemic scored development Test (PPVT) for language; Child worse than the pre-pandemic cohort in three indicators Behavior Checklist to measure out of four developmental areas. Observed social-behavioral skills; Battelle drop in general child development indicators, Developmental Inventory test for language development and social-emotional general development; and Hearts development. No difference in executive and Flowers Dimensional Stroop function. Task test for executive function Colombia 4 National Math, SABER-11 2019 and 2020 test G11 Inequality based on the characteristics of Reading, scores students increased significantly during the Science, pandemic, particularly inequality according Social to the possession of a computer or access to Studies and Internet and according to the type of school English (public or private). Mexico 5 (*) Sub- Math and Two household surveys (citizen- 10- to Learning losses occurred at all ages and SES national: Reading led assessments) from 2018 15- year-olds groups analyzed. Losses in reading were higher Southern and 2021. The “Medición for low SES group and early ages (10 and 11 Mexico Independiente de Aprendizaje” years old). For example, at age 10, for reading (Campeche (MIA) tool was applied. comprehension, scores decreased by 25% in and Yucatán) low SES groups and 15% in high SES groups. (*) The study compares the results of two household surveys that used the same tools to measure fundamental learning. For reading and mathematics, the MIA tool was applied. The first survey was conducted in November 2018 in Yucatán, Quintana Roo and Campeche. The second survey was implemented in May 2021 in Cam- peche and Yucatán. For the study, only participants between 10 and 15 years old who lived in the states of Campeche and Yucatán were selected. A socio-economic status index was generated to improve sample comparability. Sources: (1) São Paulo Education Secretariat (2021); (2) Cossi et al. (2021); (3) Abufhele et al. (2021); (4) García et al. (2021); and (5) Hevia et al. (2022). 115 https://guanajuato.lasillarota.com/estados/son-70-mil-estudiantes-los-que-desertaron-para-trabajar-en-guanajuato/633201. The impact of the COVID-19 pandemic on learning outcomes and skills development 78 Two Years After | Saving a Generation Table 3.2. Unequal learning losses: Increase in learning losses, by vulnerability characteristic Brazil 1 Chile 2 Colombia 3 Ethiopia 4 India 5 Kenya 6 Mexico 7 Pakistan 8 South Africa 9   Age/Grade Early Grades ✚ ✚ ✚ ✚ ✚ ✚ Socioeconomic Status Lower SES ✚ ✚ Gender Girls ✚ ✚ ✚ ✚ Location Rural ✚ ✚ Other areas ✚ ✚ Source: Own elaboration with information from The World Bank et al. (2021). (1) Brazil: São Paulo Education Secretariat (2021); (2) Chile: Abufhele et al. (2021); (3) Colombia: García et al. (2021); (4) Ethiopia: Kim et al. (2021); (5) India: Pratham (2021); (6) Kenya: Whizz Education (2021); (7) Mexico: Hevia et al. (2022); (8) Pakistan: ITA (2021); (9) South Africa: Ardington et al. (2021). Note: LAC region countries are highlighted in orange. Figure 3.14. Mexico: Learning loss in Mathematics, by SES level, 2019 vs. 2021 4.00 0% 3.54 3.50 3.34 3.21 3.19 -5% 3.00 2.66 -10% 2.5 2.36 2.50 2.17 Mean Math -15% Percent 2.00 -20% 1.50 -25.15% -24.86% -25% 1.00 -26.02% 0.50 -30% -32.40% 0.00 -35% Lower SES Middle-Low Middle-High High SES 2019 2021 % change Source: Hevia et al (2022). Note: Comparison of the mathematics sum means between 2019 and 2021 by SES level. The math tool includes five items (identification of numbers, addition, sub- traction, division, and problem-solving), each item has a score of 1, the total score sums individual item scores for a total ranging from 0 to 5. On average, there are greater losses for lower-SES stu- the highest SES group, or a decrease of 32% versus 25% dents, which could result in a deepening of the pre-ex- in mean scores (Figure 3.14). A significant difference was isting learning gaps between wealthier and poorer also observed in the reading learning loss. Pakistan’s poor- students. Results from Mexico reveal greater learning est children suffered absolute losses in learning and are losses in math for lower SES students: a gap in the mean far behind expected learning, while richer students re- math score of 1.04 in the lowest SES group and 0.88 in mained roughly in line with expectations116. Recent survey 116 While digital inequalities existed before the pandemic, children in poorer households found it disproportionately harder to access learning resources once technological means became the primary or only means to do schooling during school closures. There is a link between wealth, access to technology and learning outcomes of children from lower wealth quartiles. The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 79 Figure 3.15. Brazil: Share of 6/7-years-old children who cannot read and write, by income level (%), 2019 vs. 2021 (a) Children in the Bottom 25% (Poor) (b) Children in the Top 25% (Rich) 70% 30% 26% 60% 60% 25% 52% 51% 21% 50% 47% 44% 20% 18% 18% 40% 38% 16% 15% 12% 30% 10% 20% 5% 10% 0% 0% Brazil North Northeast Brazil North Northeast 2019 2021 Source: PNAD-C data (3rd quarter). Note: Poor children = lowest quartile of per capita income, PNAD-C (3rd quarter). Rich Children = highest quartile of per capita income, PNAD-C (3rd quarter). data from Brazil117, by comparing the percentage of 6- or period. In South Africa, learning losses for girls were 20 7-year-olds who cannot read and write in 2019 and 2021 percent and 27 percent higher than boys in local language by region (North and Northeast) and by socioeconomic and English reading. Girls lost 9 words per minute in read- group, confirm not only an increase in the share of chil- ing speed, versus 6 words per minute for boys, increasing dren not being able to read and write but also that this the gap. Further evidence at the country level is needed increase was especially strong in the bottom 25% students to adequately target strategies to recover learning losses, in the poor Northeast, illustrating the compounded effect particularly for vulnerable groups. Mexico also evidences of poverty and location (Figure 3.15). larger learning losses for girls than boys between 2019 and 2021 (among low-income households for reading and There is limited evidence of differences in learning loss- math, and in reading among high-income households). es by geographic location, although when disaggregat- ing by gender, girls seem to have been the most impact- ed. Notwithstanding the above, evidence remains limited on the impact of geography. In Ethiopia, learning levels of 3.2 The impact on transferable skills primary students are considerably below what would have been expected if they continued the pre-pandemic trajec- 3.2.1 Transferable skills in LAC tory. For students in urban areas, learning progressed at prior to the pandemic less than half of the speed that would have been expected if they continued at the same pace. Progress was even low- Transferable skills are those skills that allow children er for rural students whose learning progressed by one- and young people to become agile, adaptive learners third of the normal learning speed. The opposite is true and citizens equipped to navigate personal, academ- for Colombia, which shows disparities between rural and ic, social, and economic challenges.118 This definition, urban institutions- in favor of rural schools-, with a gap proposed by UNICEF's Global Framework on Transferable close to 12 points. Colombia also presents a gender gap, Skills, recognizes that transferable skills are needed to with lower average test scores (SABER-11) for women than adapt to various life contexts and which individuals can for men. However, this has been a persistent gap, and it potentially transfer to different work and social environ- has remained between 5 and 7 points over the 2019-2021 ments. Transferable skills include cognitive, social, and 117 Using PNAD-C data (3rd quarter). 118 UNICEF (2019a). The impact of the COVID-19 pandemic on learning outcomes and skills development 80 Two Years After | Saving a Generation emotional functions and can be organized in four dimen- Figure 3.16. Core transferable skills and dimensions sions,119 according to the result that its development and acquisition generates: SOCIAL DIMENSION COGNITIVE (Learning DIMENSION •  Cognitive/learning dimension (learning to know): in- to live ation Creativ (Learning ticip ity together) Par to know) cludes cognitive and metacognitive tools necessary for Cri hy t at men di t to develop new skills and to ensure the acquisition of e v ica rsity ma Emp g l e new knowledge. It implies mastery of the instruments na of knowledge – thinking, analyzing, focusing - instead ACTIVE Prob makin t LEARNING c g Respe of simply acquiring information. It includes skills such CITIZENSHIP lem as problem-solving, critical thinking, and creativity. Commun ti o n PERSONAL •  Instrumental/employability dimension (learning to EMPLOYABILITY p era EMPOWERMENT icati do): considers how to support children and young peo- Co o on ple to put into practice what they have learned, funda- n R e o lv ti o ie sil i n n ce t s ia mentally in the world of work and entrepreneurship. It g go Ne Se on- includes skills such as cooperation, negotiation, and s o l v l f- D e cisi g n INDIVIDUAL ing Thi nki INSTRUMENTAL decision-making. DIMENSION DIMENSION (Learning to be) (Learning to do) •  Individual/personal empowerment dimension (learning to be): refers to skills that allow self-realization and per- Source: UNICEF (2021e). sonal growth and foster self-empowerment and a sense of agency. This dimension is essential for personal pro- tection and violence prevention, so it must be seen as an to accumulate skills in the future, including foundational enabler of the other dimensions. It includes skills such ones.122 Some transferable skills are also associated with as communication, resilience, and self-management. greater well-being and better social interactions, by im- proving mental health, and reducing risk behaviors such •  Social dimension/active citizenship (learning to live as drug and alcohol use, crime and violence.123 together): ethical dimension that promotes individu- als to act based on human rights, consistent with dem- Measuring transferable skills presents great challeng- ocratic values, social justice and respect, for which it es.124 The lack of a commonly accepted taxonomy, the diffi- constitutes the ethical basis of the other dimensions. culty of directly observing a transferable skill, the suscepti- Includes skills for active citizenship such as respect bility of self-report - the most widely used tool to measure for diversity, empathy, and participation. transferable skills - to present measurement errors as bias- es, the lack of clarity of cross-cultural validity of the instru- Transferable skills are essential for personal, academ- ments, and the lack of understanding of what constitutes ic, social, and economic development (see Figure 3.16). the performance level of each transferable skill, among The literature on transferable skills shows that some of others, makes measuring transferable skills a complex task. these skills generate improvements in learning, wages and productivity, effects that in some cases may be equiv- Despite these challenges, initiatives to measure and alent or even greater than those of foundational skills120, monitor transferable skills were implemented in some and are highly valued in the world of work.121 Furthermore, LAC countries prior to the pandemic. For instance, on the the acquisition of transferable skills reinforces the ability social dimension, the International Civic and Citizenship 119 Delors et al. (1996). 120 Heckman et al. (2006); OECD (2015). 121 Bassi et al. (2012); Carneiro et al. (2007); Cunningham and Villaseñor (2016); Duckworth and Seligman (2005); Duckworth et al. (2007); Durlak et al. (2011); Heckman et al. (2006); Heckman and Kautz (2013); OECD (2015); Ospino (2019); Prada et al. (2019). 122 Cunha and Heckman (2007), (2008); Cunha, et al. (2006), Heckman and Kautz (2013), Busso et al. (2017). 123 Carneiro et al. (2007), Case and Deaton (2017), Durlak et al. (2011), Friedman et al. (2010), Heckman and Kautz (2012), Heckman et al. (2006), Heckman and Rubinstein (2001), Herrera et al. (2015), Lopes et al. (2004), (2011), Malouff et al. (2005), Martins et al. (2010), OECD (2015), Sánchez-Álvarez et al. (2016). 124 For more information refer to: Blyth et al. (2018), Galloway et al. (2017), Hoskins and Liu (2019), Jones et al. (2019), (2016), Lippman et al. (2014), and van de Vijver and Tanzer (2004). The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 81 Education Study (ICCS) follows students’ knowledge and un- Figure 3.17. PISA 2018 index of student cooperation derstanding of civic education and citizenship, with Chile, for LAC countries Colombia, Mexico, Peru, and the Dominican Republic par- 0.10 ticipation in 2016. PISA has measured learning, personal empowerment, and employability, alongside transversal 0.00 skills such as collaborative problem-solving and creative thinking, with over 9 LAC participating countries. The OECD -0.10 Survey on Social and Emotional Skills (SSES), where 10 cit- -0.20 ies participated, two of them from Colombia, assesses five transferable skills domains (task performance, emotional -0.30 regulation, collaboration, open-mindedness, and engaging with others). At a regional level, ERCE 2019 measured for -0.40 the first time three transferable skills - empathy, self-man- agement and openness to diversity – in 16 LAC countries. -0.50 At a national level, several countries in the region, including Argentina, Chile, Colombia, El Salvador, Mexico, and Peru, -0.60 have incorporated tools to assess specific transferable skills. Peru OECD average Mexico Dominican Republic Colombia Panama Costa Rica Chile Uruguay Brazil Argentina Education systems in the region have also put emphasis on developing students’ transferable skills by embed- ding them into the general curriculum. UNESCO's LLECE recently reviewed the inclusion of transferable skills re- Note: Positive values in this index mean that students perceived their peers to lated to education for sustainable development and glob- co-operate to a greater extent than the average student across OECD countries. al citizenship education in the national curriculum of 19 Source: Own elaboration on the basis of OECD (2019). countries in the region. The evidence shows that transfer- able skills such as participation, collaboration and critical thinking are included in the curricula of at least 16 coun- to help their family or was done without bad intentions. tries. Other transferable skills frequently included are re- The PISA 2018 results show that in all LAC participating spect, democracy, dialogue, diversity, citizenship, identi- countries, except for Peru, students perceive lower levels ty, human rights, solidarity, and responsibility, evidencing of cooperation from their peers relative to the perception the relevance given to transferable skills in the education- in OECD countries (Figure 3.17). LAC students also exhibit al systems of the region125,126. less self-efficacy127than OECD peers. Similarly, PISA 2015 evaluated collaborative problem-solving, in which LAC Some data suggests that the pre-pandemic learning also displays a lag with respect to the average levels of the crisis in LAC replicates into transferable skills. Accord- OECD and PISA participating countries.128 ing to ICCS 2016, the 5 countries with lower citizenship knowledge were the 5 LAC participating countries, and Furthermore, there are great inequities in the acqui- only Colombia and Mexico experienced improvements be- sition of transferable skills between students. Accord- tween 2009 and 2016. The study shows that two-thirds of ing to the SSES, socio-economic status, migration back- students in the LAC participating countries justify a dic- ground, gender, and age lead to differences in skill levels. tatorial regime when it brings order and security or eco- For example, in Bogotá, students with higher socio-eco- nomic benefits and most students also find justifications nomic background present higher levels of every social for breaking the law when, it is, for example, the only way and emotional skill, especially responsibility, persistence, 125 UNESCO-OREALC (2021a). 126 UNESCO-OREALC (2021c). 127 Sometimes the term self-efficacy is confused with the term resilience. Despite being related, they are not the same. Self-efficacy is the extent to which individu- als believe in their own ability to engage in certain activities and perform specific tasks, especially when facing adverse circumstances (Bandura, 1977). Resilience, on the other hand, shall be understood, in general terms, as the constructive, personal ability to navigate changing – and especially challenging - circumstances successfully (American Psychological Association’s Dictionary, see https://dictionary.apa.org/resilience). Resilient learners adjust positively to change, manage un- certainty, and respond to shocks. This starts with the student’s internal world, including emotional well-being, self-efficacy, critical thinking, and growth mindset (Chernyshenko et al., 2018). 128 Bos et al. (2017). The impact of the COVID-19 pandemic on learning outcomes and skills development 82 Two Years After | Saving a Generation Figure 3.18. Chile: Share of students reporting socio-emotional learning at school, by grade (%), 2020 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 4 5 6 7 8 1 2 3 4 Primary Secondary Express emotions Empathy with others emotions Self-regulation Source: Chile’s Agency for the Quality of Education (2021). co-operation, sociability and achievement motivation. The surveys to school principals and parents in 2021 reported ERCE 2019 results are in the same line, with socio-eco- that 20% of parents of primary school children highlight- nomic status and gender leading to differences in empa- ed higher difficulties in communication of their children, thy, self-management, and openness to diversity in almost and 10% that they lagged in socialization, aspects key to all countries. The results also show strong school effects developing social and individual transferable skills.129 on the acquisition of these transferable skills. 3.2.2 The impact of the pandemic on transferable skills 3.3 The implications of the pandemic for digital skills Students returning to school have been deeply impact- ed by the isolation and stress caused by the pandem- 3.3.1 Digital skills and “ICT ic, with expected impacts on the acquisition of trans- preparedness” prior to the pandemic ferable skills. There is still very limited evidence of the impact of the pandemic on transferable skills, but a cou- As shown in Chapter 1, many of the remote learning ple of examples point to concerns that those skills may initiatives provided by governments during school clo- have been negatively impacted. The 2021 Comprehen- sures were based on digital tools and platforms that sive Learning Diagnostics (Diagnóstico Integral de Apren- assumed previous digital skills proficiency. The digi- dizajes, DIA) developed by the Education Quality Agency tal divide in the region not only relates to infrastructure, of Chile (Agencia de Calidad de la Educación) assessed in a devices, and connectivity, but also to digital literacy and comprehensive fashion learning and socio-emotional sta- skills levels. Governments, schools, teachers, and students tus of students returning to schools. 90% of the surveyed had to quickly adopt educational technologies, and avail- students in the last two years of secondary education re- able information suggests that weak digital skills might sponded that the lack of face-to-face education affected have hindered remote education efforts. There is evidence them negatively in their learning process. Around 44% that both students and teachers lacked digital literacy and of secondary schools’ students stated that they were not skills prior to the pandemic. taught emotional regulation or empathy at their schools, hindering the development of social and individual trans- The concept of “digital literacy” in educational set- ferable skills (Figure 3.18). In the Dominican Republic, tings is an umbrella concept that covers a wide range 129 WBG (2022b). The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 83 Figure 3.19. Share of students in LAC countries whose school principals agreed with the statement “Teachers have the necessary technical and pedagogical skills to integrate digital devices in instruction” (%), 2018 76.5% 72.2% 64.6% 62.2% 59.4% 55.5% 54.5% 54.1% 50.6% 49.5% 40.9% Mexico Panama OECD average Chile D.Rep. Colombia Costa Rica Peru Brazil Uruguay Argentina Source: Own elaboration on the basis of OECD (2020a). of skills and tasks,going from the basic technical skills Initiatives to measure children and teenagers’ digital that are needed to interact with teachers through mo- skills were implemented in some LAC countries. For in- bile apps to receive and send information (for instance, stance, Brazil, Chile, Costa Rica, and Uruguay participated in for the WhatsApp exchanges widely used during school different waves of the Kids Online project, a survey that col- lockdowns) to the advanced skills needed to use adap- lect data on behavior and attitudes of children online, with tive learning platforms that allow for personalized in- special focus on safe use of Internet133. Chile and Uruguay struction, or to interact with software that facilitates syn- participated in the International Computer and Information chronic interactions between students and teachers130. Literacy Study (ICILS), that focuses on secondary education Different strategies deployed by governments during students. ICILS 2018 findings showed interesting findings, the lockdowns required simpler or more complex digital that questioned the idea of “digital natives”: more than one abilities131, and very limited diagnostics on pre-existing third of students underperformed in digital competence. Girls levels of such abilities were available to guide the policy outperformed boys in all participating countries, and, finally, decisions132. results confirmed the correlation of lower socioeconomic sta- tus and low performance in digital skills and computational There has been great improvement in national assess- thinking, suggesting a risk of a persistent digital divide134. ments of learning outcomes in LAC in recent years, but not a similar progress in data collection about students’ Teachers as well lacked digital skills. Data collected by digital literacy. Younger generations are often considered OECD in 2018 showed that, on average, one third of the “digital natives”, but data on specific digital competencies students included in PISA attended schools where teach- is scarce, and available information suggests that digi- ers lacked technical and pedagogical skills to effectively tal literacy is far from universal in the region, and even incorporate the technology needed for distance learning. though the growth of remote learning might have had a Most of LAC countries were in a worse position than the positive impact for children and youth digital skills, ineq- OECD average regarding their teachers’ digital education uities in the coverage of connectivity and digital devices readiness, according to their school principals’ percep- might have increased gaps in such skills. tion (Figure 3.19). Evidence (even if limited) on teachers’ 130 UNICEF (2020). 131 Muñoz-Najar et al. (2021) p. 48. 132 UNESCO (2019). 133 Trucco and Palma (2020). 134 European Commission (2019). The impact of the COVID-19 pandemic on learning outcomes and skills development 84 Two Years After | Saving a Generation digital abilities, defined beyond basic literacy, as the cog- Figure 3.20. Share of teachers by level of digital nitive competencies required to solve complex problems skills, by region (%), 2020 in a digital environment, suggests that deficits exist. An 60% analysis of teachers’ skills published in 2020, based on the Program for the International Assessment of Adult Com- 49% 50% petencies (PIAAC) databases, showed that 46 percent of the teachers of the four LAC countries included (Chile, Ec- 39% 40% 39% 40% uador, Mexico, and Peru) did not reach an adequate level of digital skills, compared to 12 percent of the teachers in 30% OECD135, as shown in Figure 3.20. 20% Before the pandemic, digital training offer for pre-ser- 13% vice teacher was often very basic and not systematic 9% 10% 7% in the region, which would likely explain these gaps. Pre-service teacher training, even including elements for 1% 0% the development of digital skills, usually does not involve No skills Below level 1 Level 1 Levels 2 and 3 advanced training for incorporating technology into peda- LAC (4 countries) OECD (15 countries) gogy (such as the use of adaptive learning technologies, the Source: Own elaboration on the basis of Estrada and Lombardi (2020), using integration of computational thinking across areas, the cre- PIACC databases. Includes teachers aged 22 to 65 years old, including all ation of digital content, or issues related to safety in digital educational levels. environments). For instance, Uruguay was the only country Note: Levels 1-3 correspond to different levels of “problem solving in techno- logy-rich environments”. LAC countries include Ecuador, Chile, Mexico, and that before the pandemic had developed all the basic condi- Peru. tions for digital education (school connectivity, digital plat- forms, virtual tutoring, digital resources packs and digital repository136). However, even in Uruguay, universal digital before, and 83 percent felt little or not at all skilled to do skills training for teachers was not in place: even though the it.137 As Muñoz-Najar and his colleagues suggest, there national EduTech program Plan Ceibal offers a wide range is no robust evidence on the effectiveness of the teach- of professional development tools, not only on the specif- er training programs, and anxiety and stress reported by ic abilities needed to use the digital tools and platforms, teachers might have jeopardize their impact138. but also on innovative educational trends, the offer is still not incorporated into pre-service programs, and no digital 3.3.2 Effects of the pandemic training is required to become a teacher. on digital skills Survey data suggest that insufficient prior use of digital LAC countries were proactive in offering support for skills, and limited effectiveness of training programs, teachers’ digital skills development during the pan- constrained readiness levels and created high level of demic. In some cases, the education authorities strength- anxiety of teachers for remote education. Even before ened and extended tools and strategies that were already the pandemic, in some LAC countries there were percep- in place, but in some other cases the support was creat- tions, among school staff, that teachers lacked sufficient ed as a response to the pressing remote teaching needs. time to develop their educational digital skills. Survey Figure 3.21 summarizes some good practices implement- data collected in 2020 in Brazil showed that 67 percent ed in Uruguay, Costa Rica, and Peru, were toolboxes and of teachers felt anxiety about remote teaching, 88 percent dedicated platforms were made available for teachers, declared that they never implemented remote learning students, and families. Not only national governments 135 Estrada and Lombardi (2020). 136 Rieble-Aubourg and Viteri (2020). 137 Muñoz-Najar et al. (2021). 138 Muñoz-Najar et al. (2021, p. 28). The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 85 Figure 3.21. Examples of digital toolkits with teaching resources DIGITAL TOOLKIT Costa Rica Guide for autonomous Training –Aula virtual– pedagogical tasks Peru Massive on-line –Aprendo en casa– learning courses Uruguay Guidelines for remote Virtual training Forums –CREA– teaching Source: Barron-Rodriguez et al. (2021a). Even though CREA had been available for teachers for many years before the pandemic, while 40% of grade 6 and 25% of grade 3 teachers were using it in 2017 (Carbajal et al., 2022), these percentages rose to 95% in 2020, and 92% of the teachers were satisfied with the training offer (Ripani, 2020). provided support for teachers digital training.139 It is not in hybrid and face-to-face learning models. Students may clear however yet how effectives have been these new also have become more proficient in making effective use of tools and strategies, especially considering the challenges those digital tools. School closures also helped to show the faced by the remote learning response. crucial role of parents and families in the learning process, and the role and use of technology in facilitating a close One of the consequences of the pandemic on education interaction, experiences and abilities that can indeed also was that, more than ever before, schools were pushed continue playing a role in normal times. Also, the pandemic to adopt tools that implied a certain degree of digital pushed knowledge and debate on the advantages and lim- ability. Thus, a very relevant question for the post-COVID its of learning platforms, remote learning, hybrid models, scenario is whether the diverse kinds of remote educa- and remote assessment strategies among policy makers tion strategies had any positive effect on teachers and and opinion leaders, discussions that might accelerate in- students’ digital skills. The answer cannot be definitive at novation at faster pace in the medium run141. this stage, as the first assessments of learning that coun- tries implemented were, sensibly, mostly focused on foun- dational skills such as reading and mathematics, but there are indications of potential positive trends. School closures 3.4 The impact on psychosocial showed, at large scale, that the kind of digital skills that health and well-being teachers need are not limited to the use of technology, but also extend to particular types of pedagogical and commu- While not skills per se, psychosocial health and well-be- nicational skills, as “simply replicating a lesson designed for ing play a critical role in supporting the acquisition of an in-person class is ineffective for a remote audience”140. all other skills. Psychosocial health and well-being are Digital skills not only involve knowledge of digital tools, but critical to ensure students can foster an adequate set of judgement on why and how to incorporate them to max- transferable skills and are equipped with critical functions imize impact. The higher exposure to these skills during to be ready to acquire the fundamentals of the learning the pandemic could be very helpful in the longer-term in process. They also contribute to increasing attendance enhancing an effective use of technology in the classroom and reducing dropouts. While we do not have separate 139 A rather interesting experience, not implemented by a national ministry but by a civil society organization, the Varkey Foundation, was Comunidad Atenea, an online community of practice where teachers from all over the region could connect, share resources, and create collaborative contents. Some of the features of Co- munidad Atenea are similar to social networks, including the possibility of bilateral interactions, and private messaging. Within the platform, online training courses were offered, for teachers to enhance their digital abilities. The relevance of the communities of practice during the pandemic was higher than ever, not only allow- ing the teachers to support each other in the absence of the normal school community of peers, but also communicating and multiplying innovative experiences in the use of technology for education (IAD, 2021a). 140 Wilichowski, T. and Cobo, C. (2020). 141 Zancajo et al. (2022). The impact of the COVID-19 pandemic on learning outcomes and skills development 86 Two Years After | Saving a Generation Figure 3.22. Argentina: Share of youth reporting feeling distressed, scared, or depressed by the uncertainty of the pandemic (%), April 2020 – October 2021 80% 70% 18% 60% 11% 50% 12% 25% 6% 40% 25% 30% 14% 12% 23% 20% 8% 33% 27% 24% 10% 16% 16% 0% Apr-20 Jul-20 Oct-20 May-21 Oct-21 Distressed Scared Depressed Source: UNICEF (2021a). evidence of these dimensions before the pandemic, we do Docente, ENDO) reported that the top 3 difficulties faced by have emerging evidence of a mental health crisis within a teachers in 2020 were supporting student´s socio-emotion- crisis, which will also require urgent attention. al problems, communicating with and motivating students. There is strong evidence of a deterioration of mental Teachers have also seen their personal, emotional, and health and well-being of children and youth across edu- professional life disrupted during the pandemic. Togeth- cation levels142. A systematic review of 36 studies from 11 er with children and parents, teachers had to rapidly adapt countries, including Brazil from the LAC region, finds that to new modalities of teaching and to the challenges of re- school closures and lockdowns during the first COVID-19 mote working during the pandemic, including childcare, wave were associated with adverse mental health symp- domestic chores and taking care of family or parents. The toms (such as distress and anxiety) and health behaviors results of ENDO 2020 in Peru show that 16.2% of teach- (such as higher screen time and lower physical activity) ers reported symptoms of depression, 27.3% symptoms of among children and adolescents.143 For younger students, anxiety and over 60% reported stress during 2020. As il- surveys of caregivers across the region have consistently lustrated above, switch to remote learning in a context of documented shifting behaviors, with 61% of children dis- limited readiness also generated high anxiety in teachers. playing at least one symptom of mental distress.144 At the higher education level, 36% and 65% of 18–29-year-olds in Mexico and Chile, respectively, report their mental health and well-being, or that of a household member, has been 3.5 Key takeaway messages affected by the pandemic and crisis.145 In Argentina, high frequency surveys to teenagers from April 2020 to Octo- The LAC region was already suffering from a serious ber 2021 find high levels of self-reported pandemic stress learning crisis before the COVID-19 outbreak. Unweight- and depression (Figure 3.22), with over 29% of adolescents ed results from ERCE confirm that in 2019, on average, al- not looking forward to returning to school. Data from the most half of 3rd graders did not reach the minimum profi- National Survey of Teachers in Peru (Encuesta Nacional ciency level (MPL) in reading and more than two-thirds did 142 For a comprehensive review of the impact of the pandemic on children’s well-being, see UNICEF (2021f). For a treatment of the impact of the pandemic of youth’s well-being, see IDB (2022), chapter 2. 143 Racine et al. (2021). 144 Näslund-Hadley et al. (2020). 145 OECD Secretariat estimates based on the OECD Risks That Matter 2020 survey, http://oe.cd/RTM. The impact of the COVID-19 pandemic on learning outcomes and skills development Two Years After | Saving a Generation 87 not reach it by 6th grade. Results also showed the highly The pandemic has also affected transferable skills, unequal profile of the region, with wide disparities among while showing the importance of digital skills. Some countries and a pervasive persistence of socioeconomic data suggests that the pre-pandemic learning crisis in LAC learning gaps. Disparities in learning outcomes were also replicates into transferable skills. There are also great in- largely explained by differences within classrooms. equities in the acquisition of transferable skills between students. Very incipient evidence shows that these skills The COVID-19 pandemic has deepened the crisis by se- may have been affected, especially by the isolation related verely impacting learning, with a deeper impact on the to school closures. At the same time, the pandemic also earlier grades and lower-income students. According to revealed gaps in digital skills, while, in a bit of a silver recently updated simulations, the average number of LAYS lining, providing some opportunity for enhancing those lost due to the pandemic is expected to be about 1.5 years, skills. One of the pandemic consequences on education under an intermediate scenario, translating into a decline of was that, more than ever before, education systems and 12% in the projected annual earnings of the average LAC stu- schools were pushed to adopt tools that implied a certain dent at school today. The proportion of 3rd and 6th graders not degree of digital ability enhancing exposure to these skills able to adequately understand and interpret a text of moder- for teachers, students, and parents, while strengthening ate length could have increased, respectively, from weighted tools and strategies to develop digital skills. averages of 36 percent to 50 percent and 62 percent to 82 percent. Initial data on actual losses confirm these trends. Finally, recent evidence on the post-pandemic points Overall, simulations and the available evidence indicate to an important deterioration of well-being of children that, on average, early grades, younger children, and chil- and youth across education levels. This will hinder the dren from lower SES have been disproportionally affected acquisition of transferable skills and readiness to learn, as by greater learning losses. In combination with the emerg- well as attendance outcomes. ing increasing number and share of out of school children in the 6-14 age range, this lays the ground for a major genera- The gravity of the crisis requires a prompt and compre- tional crisis. This is especially worrisome as, in turn, cumu- hensive response. The next chapters document actions lative learning gaps have the potential to further increase taken so far by LAC countries and the potential and urgen- dropouts of this younger cohort when it becomes older. cy for more. 88 Two Years After | Saving a Generation PHOTO BY: © STUDIOLASKA / SHUTTERSTOCK.COM Chapter 4 Recovering and accelerating: Returning to schooling Recovering and accelerating: Returning to schooling Two Years After | Saving a Generation 89 T he COVID-19 pandemic has generated a negative shock in the education sector worldwide. Chap- ters 2 and 3 discussed the impact that prolonged school adequate and efficiently used resources, to make their im- plementation possible. closures had on the two critical educational outcomes As countries recover, they should also aim at acceler- schooling and learning. They also analyzed the challenges ating. Given the need to recover, in principle, in the same that the pandemic brought about transferable skills and number of schooling years, children and youth will need to digital skills, as well as well-being. Both chapters con- be put on an “accelerated pathway” to learning recovery, firmed that the impact was disproportionately higher for which may not only help recover in the short-term but also vulnerable groups, i.e. students from more disadvantaged achieve higher levels of learning than before the pandem- socio-economic backgrounds, and, at least at this stage, ic, a goal very much needed in the region given the prior for the younger cohort. But what can and should be done low learning levels. Countries can also build on recovery with this somber diagnostic? If urgent action is not taken, efforts to make important strides on curricula and peda- an entire generation may suffer profound and long-lasting gogical approaches. Policies to foster return to schooling consequences on their human capital accumulation. may not only help consolidate the decline in out of school rates but also accelerate it. Recovery must focus on the essentials: returning to schooling and recovering from learning losses (Fig- This chapter delves into the “return to schooling” agen- ure 4.1).146 The “return to schooling” aims at ensuring da. More than two years have passed since the start of that all children who were attending education prior to the pandemic, and countries in the region have been pro- the pandemic can return to school, and the decline in out gressively reopening their school systems and prioritizing of school rates continues. Once students are back in the a return to face-to-face classes for students. Recent data classrooms, actions also need to be taken to offset the indicates, however, that still roughly 12 million (roughly 10 learning losses of the past two years (“recovering from percent of) students in the region are currently not benefit- learning losses”). To be successful, this double schooling ting from face-to-face classes. Additionally, risks of schools and learning agenda will require addressing the psycho- closing again are real depending on the evolution of the social challenges triggered by the pandemic, as well as pandemic. This chapter discusses the key immediate chal- continuing to address the digital divides that have been lenges that need to be borne in mind and addressed to ap- hampering the response. Embedded in all these priorities proach the “return to schooling” problem: (i) the reopening is the support to teachers and principals, and ensuring challenge; and (ii) the enrollment and drop-out challenge. Figure 4.1. The road to recovery and acceleration in LAC RECOVERY FROM LEARNING RETURN TO SCHOOLING LOSSES AND ACCELERATION RECOVERING Safely Re-Enrolling Preventing Consolidating Assessing Implementing reopen dropouts the curriculum student’s learning recovery schools learning level programs • Addressing psychosocial needs • Addressing the digital gap 146 This framework is well aligned with the RAPID framework recently proposed by UNICEF, UNESCO, and WBG (2022). The framework is based around 5 pillars: (i) Reach all children and adolescents that should be at school and retain them in school; (ii) Assess learning levels; (iii) Prioritize teaching the fundamentals, namely critical foundational (literacy and numeracy) and problem-solving skills (critical thinking, innovation); (iv) Increase catch-up learning and progress beyond what was lost; and (v) Develop psychosocial health and well-being. Recovering and accelerating: Returning to schooling 90 Two Years After | Saving a Generation 4.1 Reopening schools However, prolonged closures do negatively impact children’s learning, health, and well-being especially Throughout most of the first year of the pandemic, for vulnerable groups. As shown in the previous chap- schools in the region remained closed or mostly closed, ter, children’s ability to read, write and do basic math as a health measure to control the spread of COVID-19 has suffered, as has the development of their skills to and safeguard students and teachers’ health. This left reach their future potential. Many studies have consis- more than 170 million students in LAC out of the possibil- tently shown that children’s health, development, safe- ity of in-person education, many of whom suffered from ty, and well-being also took a hit154. And, as was em- the disconnection that forced distance learning strategies phasized repeatedly before, the most vulnerable boys entailed especially for the most disadvantaged popula- and girls bore the brunt of these impacts, intensifying tions147. During the second half of 2021, as health condi- pre-existing gaps with their better-off counterparts. The tions improved, many countries started to progressively urgency of redressing this situation for those countries reopen schools and transitioned from fully remote to hy- in LAC whose systems are not fully open (see section 1.2 brid learning models. At the beginning of 2022, two years above) is to understand the criticality of evidence-based after the initial outbreak of the pandemic, there has been COVID-19 policymaking in schools155. And any strategy a larger-scale reopening of school systems in the region that intends to address the “return to schooling” chal- and a larger proportion of students have returned to face- lenge must design a clear roadmap that focuses on the to-face classes. As a result of the heterogeneity of school top priorities for the short term. systems in the region, though, different countries remain at different stages of this recovery process. And it is also The first priority in the recovery phase is the safe essential to ensure that school reopening is sustainable reopening of schools, to ensure that all children re- through the right measures and policies. turn to face-to-face education as soon as possible and for good. School reopening should be a top priority Evidence collected during these two years shows that for those school systems that remain closed or mostly schools are not an important source of COVID-19 trans- closed, especially. And lessons from existing experiences mission148. In fact, in-person schooling does not appear to of school reopening should be built on to complete this be the main driver of infection spikes, children in school process and/or make sure it is sustainable. As highlight- do not appear to be exposed to higher risks of infection ed above, two reasons are the key rationale for this ur- compared to when not in school when mitigation measures gent strategy: (i) COVID-19 poses limited health risks for are in place, and school staff also do not appear to be at younger populations; and (ii) foundational skills are not a higher risk compared to the general population149. Mul- adequately developed at home. While it is not possible tiple studies have shown that transmission within school to completely eliminate the risk of disease transmission settings is typically lower than community transmission in schools, evidence from reopened schools indicates when prevention strategies are in place for schools150. Fur- that simple mitigation measures like masking, social thermore, with adherence to basic mitigation strategies151, distancing and ventilation are effective in substantially within-school transition of COVID-19 can be interrupted, reducing transition to students and teachers. To ensure even during a surge of community infections152,153. School a safe return, it is therefore essential to take context-ap- reopening decisions should reflect evidence-based esti- propriate health and hygiene protocols (Figure 4.2). In mates of the risks associated with reopening and the risks addition, health policies, such as the prioritization of associated with school closures. teachers in the vaccination process, should allow for a 147 UNICEF (2021b). 148 Ismail et al. (2021). 149 UNESCO and UNICEF (2020). 150 CDC (2021). 151 Prevention strategies include promoting vaccination, consistent and correct use of masks for people who are not fully vaccinated, physical distancing, screen- ing testing in schools to identify cases, improved ventilation, handwashing and staying home when sick, contact tracing and routine cleaning with disinfection. 152 Zimmerman et al. (2021a, 2021b), UNESCO and UNICEF (2020). 153 Ertem et al. (2021) use data from the United States to examine the effects of different schooling models on COVID-19 case rates by comparing in-person school- ing with hybrid or virtual models of education. Similarly, Fukumoto et al. (2021) compares COVID-19 case rates in Japanese municipalities where schools opened with similar municipalities that kept schools closed. Both studies conclude that there is not a consistent relationship between school reopening and COVID-19 case rates. 154 Skovgaard Jensen (2021); Hu et al. (2021). 155 Stuart and Dowdy (2021). Recovering and accelerating: Returning to schooling Two Years After | Saving a Generation 91 Figure 4.2. Some critical safety standards for Figure 4.3. Vaccination of the population and school reopening schools reopening rates in LAC Guarantee the good condition of the sanitary infrastructure in all schools, as well as access to water, soap, and disinfectants. 80% Implement sanitation procedures. Chile % of people fully vaccinated against COVID 19 70% Equip/repurpose/reconfigure schools to support social distancing. 60% Ecuador Argentina El Salvador Ensure communication and behavior change efforts of all 50% students related to adequate handwashing, hygiene, physical Peru Mexico distancing, correct mask use and safe water practices at schools. 40% Colombia Develop clear and easy-to-understand protocols on hygiene Belize measures and physical distancing measures. Bolivia 30% Paraguay Establish a monitoring mechanism to detect and contain 20% infections in a timely manner and protocols for isolating cases, Guatemala tracing their contacts, and quarantining exposed individuals. 10% 0% Consider teachers as a priority group for vaccinations. 0% 20% 40% 60% 80% 100% % of students bene ted by school reopening Source: UNESCO, UNICEF, WBG, World Food Programme, and UNHCR (2021); Zimmerman, Brookhart et al. (2021); Zimmerman, Akinboyo et al. (2021) Note: Vaccination rates are obtained from Our World in Data (https://ourworl- dindata.org/COVID-vaccinations). Alternative definitions of a full vaccina- tion, e.g. having been infected with SARS-CoV-2 and having 1 dose of a 2-dose protocol, are ignored to maximize comparability between countries. School prompter return to school. In fact, teacher vaccination reopening rates are obtained from the UNICEF COVID-19 Education Respon- rates have been instrumental in the school re-opening se Update 29 (October 2021) (https://www.unicef.org/lac/media/29081/file). Both variables consider data from October 2021. process around the world. To date, more than half of LAC countries have most teachers vaccinated with the sec- ond dose of the vaccine. understanding of the epidemiological risks associated A strong recovery phase requires trust amongst stake- with reopening and how to tackle them have been two holders, and governments should continue to prioritize of the potential reasons for limited reopening in certain school staff for vaccination to promote confidence in a countries. Low-income countries have struggled the most safe school reopening for in-person learning. A compar- with more expensive and coordination-intensive activi- ison of the proportion of people fully vaccinated with the ties, as well as with ensuring that even the most basic extent of school reopening (Figure 4.3) points to a gener- health mitigation measures are in place. For instance, ally positive correlation which highlights the importance 27 percent of low-income country respondents in the of vaccination for school reopening. At the same time, as National Education Response to COVID-19 School Clo- illustrated below, cases like Peru and Ecuador show that sures survey156 reported that their country implemented while vaccination rates are an important part of the equa- guidelines in all of the schools compared with the global tion, other health and hygiene factors and the extent to average of 55 percent157. This points to the importance of which they are applied and communicated are also im- clear guidelines for prioritization, flexibility and provid- portant to recover trust and build impetus for a school ing, as needed, additional training and support to prin- reopening. cipals to lead school reopening to help with the coordi- nation, operation and communication challenge that a An effective school reopening process and learn- safe school reopening process entails. A recent ad-hoc ing recovery requires prepared and well-informed training program for principals in Peru provides a good and aligned leadership. Lack of clear communication example of support to principals in a difficult reopening about the limitations of remote learning, and limited context (see Box 4.1). Aligning messages, incentives and 156 It is important to note that the UNESCO/UNICEF/WBG/OECD survey had 3 rounds. The last round was carried out in Feb-June 2021 (see details at https://tcg. uis.unesco.org/survey-education-COVID-school-closures/). 157 UNESCO, UNICEF, WBG, and OECD (2021a). Recovering and accelerating: Returning to schooling 92 Two Years After | Saving a Generation Box 4.1. Challenges for and support to school reopening in Peru Peru’s schools shut down completely as the pandemic began in March 2020. Since then, schools have been reopen- ing, but closures have been among the longest in the region. Critical constraints to school reopening have related to lack of flexibility in the implementation of school reopening plans, lack of resources for infrastructure upgrades, limited preparedness of principals and lack of alignment in the incentives of principals and regional departments. To support reopening, Peru and the World Bank have been working on a series of measures, including communication tools and advocacy, costing tools and principal training. School reopening costing tool to estimate resource needs to implement government school reopening plans. Signif- icant investment in infrastructure was needed to facilitate school reopening, since half of schools do not have basic services, including water and electricity, rising to 80% in rural areas. In 2021, the World Bank piloted a school re- opening costing tool adapted to the country context and needs to estimate the resources needed to reopen schools in rural secondary schools. Consultation meetings were held with the government to explain the objective and design of the tool and to understand the country context. Training sessions also took place. Support to school principals for safe schools reopening. Early in the reopening process, school principals, who have an important say in school reopening, received multiple sources of information and were required to work more hours without receiving any extra ad-hoc training to adapt changing protocols to the new context. They were also asked to deal with multiple tasks with limited guidance on what to prioritize. In January 2021, the World Bank started a partnership with the Ministry of Education of Peru to pilot and evaluate a training program designed by the NGO Global School Leaders (GSL) and adapted it to the local context: • The Leadership for Renewal (LFR) training program aims to improve school principals’ organizational and com- munication skills to reduce the opportunity cost of school reopening and potentially contribute to improved attendance among students whose parents receive more timely communication. • The pilot would identify modules and opportunities for improvement in the design before scaling contents to Per- uEduca, the national educational platform that provides free teaching resources. It was applied in 2005 schools: 1,123 treatment, 580 control and 302 pure control. The training process included a first-level training of Local Education Authority specialists provided by the Varkey Foundation through synchronous modules, in partnership with GSL, between June and July of 2021. Between August and September of 2021, specialists trained school prin- cipals from their jurisdiction on modules A and B (see figure below). • Evidence from focus groups organized with a set of participants highlighted that receiving training on topics such as managing well-being (A3) and tools for communication (B1) was especially helpful to support their commu- nication with teachers and parents and promote confidence in a safe reopening process. A program like LFR is a positive complement to the support and online modules countries in the region have offered their school leaders. A. School Reopening B. Technology for Learning and Management A1. Preparing the School B1. Tools for Communication A2. Preparing the Staff B2. Tools for Learning A3. Managing Well Being B3. Tools for Management A4. Community support Source: Tanaka et al. (2021) Recovering and accelerating: Returning to schooling Two Years After | Saving a Generation 93 support across schools and other levels of government Figure 4.4. Link between early shock before age six also can play an important role, as shown below in the and well-being later in life case of Chile. In the process of prioritizing school reopening, open- Test scores ing ECE and primary education institutions should be a top priority.158 We have seen in previous chapters that Mental Cognitive health ability primary education children have been the ones to suffer the most in terms of attendance and learning. It is there- fore imperative to give priority to the reopening of prima- Shock begore ry schools. At the same time, and while the evidence on Use of age 6 High ECE is scarce, we also know that pre-school children have socal school assistance graduation been deeply impacted and that any action to save this generation of children, and do it efficiently and effective- ly, needs to start from this very early age. Early childhood Criminal experiences have a profound impact on brain develop- Earnings activity ment-affecting learning, health, behavior and ultimate- ly, lifetime opportunities. There is evidence from exper- imental and quasi-experimental studies that preprimary education improves children’s skill development – both for cognitive skills like early literacy and numeracy and were closed between March and September of 2020 and social-emotional skills (like perseverance and focus). Fig- began to open partially between October and December ure 4.4 lists outcomes for which we have solid causal ev- of that same year. A total of 129 municipalities autho- idence linking early life circumstances to well-being later rized schools to partially reopen for in-person activities, in life. Preschools and childcare centers also tend to be benefitting around two million students. A recent study places where children can receive psychological support found that, on average, no systematic association was to deal with stressful home environments, stress which, found between school reopening and COVID-19 inci- as was documented, has increased during the pandem- dence or mortality in São Paulo State, up to 12 weeks ic. Reopening preschools can also help mitigate stress for after reopening159. School reopening in São Paulo State, female teachers who already must take care of students while relatively timely, did not however prevent signifi- having fallen behind. cant learning losses. A good practice in school reopen- ing in the region is the case of Chile which spearheaded LAC has had some good practices on the school re- high quality, coherent and well-coordinated initiatives, opening process, including experiences from Uru- related to safety and broader recovery, making them guay, Brazil (São Paulo) and Chile. Uruguay was the available flexibly to schools to enable them to choose first country in LAC to successfully return to face-to-face the most relevant ones to their capacity, needs and con- classes. The return to schooling was a gradual process texts. Chile may however need to put more focus on that began by opening rural schools in April of 2020. By the capacity of lower-income schools to help them ap- June 2020, almost all schools had reopened. The Health ply more of those initiatives. A recent study from Chile Administration drafted a protocol that schools should showed that the amount of time schools remained open apply, with special attention to maintaining physical dis- during the pandemic has become additional information tancing in the classrooms and the playground. Face-to- that families use when choosing a school for their chil- face attendance was not compulsory and was combined dren160, suggesting that constraints have been more on with online learning. In São Paulo State, Brazil, schools the supply-side. 158 This is not to minimize that there may also be a strong rationale to reopen higher grades to address teenagers’ socio-emotional issues, potential crime, and violence situations, and even teenagers’ pregnancies. Hence, reopening all schools should really be the key objective at this stage of the pandemic. 159 The differences in disease activity after October 2020 across municipalities that authorized schools to reopen and those that did not were small and not sta- tistically significant. See Cossi et al. (2022). 160 The study shows that the percentage of time that schools were open has a positive and significant effect on the preferences of families (expressed as a great- er demand for available spots in the school). The study concludes that the opening of schools has been valued by families and that this information has become an important reference for their decisions regarding school selection. See WBG and CEM (2022). Recovering and accelerating: Returning to schooling 94 Two Years After | Saving a Generation However, some countries in the region have faced sig- 4.2 Re-enrolling and preventing nificant challenges for school reopening, as is the case dropouts of Ecuador, El Salvador, Honduras, Panama, Peru, and Venezuela and all federal countries. Challenges with Enrollment and attendance are crucial indicators to reopening schools were many. Regrettably, constraints monitor to determine the schooling status, an essential were found, in many of these cases, in the lack of re- step to ensure learning and good educational outcomes sources to make significant investments in infrastructure in the short and long term161. As schools reopen it is es- to adapt spaces to social distancing, or more broadly, sential that all students get back to school. Additionally, simply because of lack of basic services like water and/ once in school students must attend it and not drop-out. or needed hygiene-related inputs. Other challenges were This requires attention to policies to attract children and related to lack of preparedness of principals, incoher- youth back to school and keep them in school. ence in guidelines and disconnects between schools, re- gional and/or central governments. Coordination issues As presented in chapter 2, the assessment on the im- were especially clear in federal countries as exemplified pact of the pandemic on schooling so far for the region in Box 4.2. has been mixed, but there are strong risks. The arrival of the pandemic and school closures as a containment mea- To sum up, school reopening strategies are the first sure had significant implications on school attendance most important step to be taken for those systems rates and engagement. Attendance rates decreased, es- that remain closed or mostly closed. There is substan- pecially for children and youth in more vulnerable fami- tial evidence amassed during these two years into the lies. Additionally, the 6-to-14-year-old group does seem to pandemic that clearly concludes that: (i) in-person edu- have been negatively impacted in terms of attendance by cation is the most effective education service modality; the end of 2021, at least in selected countries. At the same and (ii) school classrooms are not necessarily a vector time, in the same timeline, the 15-to-17 years old had not of COVID-19 transmission, if school health protocols are been (yet) really affected, and so overall enrollment rates applied consistently, hygiene resources provided, and dis- (in an admittedly limited country sample). Enrollment tancing observed. Hence, ensuring school reopening must rates for the 2022 academic year are however not yet be the single most important first strategy for bringing all available, and it is possible that many students did not re- students back to school. And lessons from existing experi- enroll, and there is a real risk of future drop-out, especial- ences of school reopening should be built on to complete ly of vulnerable children and youth, based on their atten- this process and/or make sure it is sustainable. Short-term dance and learning gaps, and how the economic context recommendations for school re-opening are therefore: (i) and financial constraints may evolve. For example, simula- design a school re-opening strategy based on latest evi- tions undertaken by the Ministry of Education in Chile, in- dence of the health situation locally, coherent and clear- dicated that dropout rates were expected to increase from ly owned by principals and all government levels; (ii) es- 2.6% to 5% due to the pandemic162, following decreases in tablish and fund proper safety standards for schools that attendance rates. Policies for re-enrollment — even if just could be closely observed and clearly communicated to all for the second part of 2022 and 2023 at this stage — and stakeholders; and (iii) if unable to open fully the school drop-out prevention need to be put in place. system, make an effort to prioritize specific levels of edu- cation when open partially, starting from the lower levels This assessment and the related policies can provide of education. Hybrid learning models could be a short- for some useful lessons. As was shown in chapters 1 term option to address social distancing issues in schools, and 2, governments from across the region implemented and also helpful in supporting more resilient systems in different types of policies to counteract the negative ef- the medium-term in case schools have to close again or fects of the pandemic, some of which ended up positively simply to create new opportunities for enhancing teach- affecting enrollment (directly and indirectly, see Box 2.1 ing and learning (see Box 4.3 for some useful lessons on in Chapter 2) and attendance. Consequently, those coun- hybrid learning). tries which are dealing with impending school reopening processes, and all countries generalizing reopening in the 2022 academic year, should consider building on at least 161 UNESCO, UNICEF, and WBG (2021c). 162 WBG et al. (2021). Recovering and accelerating: Returning to schooling Two Years After | Saving a Generation 95 Box 4.2. School reopening in federal countries: Challenges and lessons learned Argentina Federal Organization: Challenges: Lessons learned: 24 sub-national states, en- Consensus between national and sub-national level: In March 2021, the Na- Importance of consen- compassing 23 provinces tional Ministry of Education issued two resolutions for the nationwide return sus and promotion of and one federal district to classes for the new school year 2021. Shortly after, in April 2021, a sudden plural spaces for deci- - the Autonomous City of surge in contagions led the President to announce the suspension of in-per- sion-making with di- Buenos Aires (CABA), the son classes nationwide. The Government of CABA rejected the re-closing of fferent stakeholders, capital of the country (1). schools and disputed this new Presidential decree on two grounds: (i) dis- specialists, teachers, trict’s autonomy in education management dictated by the country’s Natio- families, and students. nal Constitution; and (ii) detailed sanitary data compiled at the district le- vel, which showed a significant deviation from national average indicators. Importance of EMIS for Compulsory education: basing school reope- The controversy was judicialized and the National Supreme Court ultimately Decentralized fully to the upheld CABA’s decision. Immediately after this legal decision was ratified, Ar- ning decision- making sub-national level of first or- gentina had simultaneously the bulk of its education system fully/mostly clo- on monitoring of sani- der (23 provinces + CABA). sed, with the education system in CABA being fully open. tary indicators at the local level. Brazil Federal Organization: Challenges: Lessons learned: 27 sub-national districts, Safe reopening of schools with adequate conditions (water, health, infrastructu- Importance of deter- comprising 26 states and re): The application of safe protocols for school reopening, and the concomitant mining funding obli- one federal district - the need for investment (e.g. small civil works, refurbishment, ventilation arrange- gations for safe school city of Brasilia, the capital ments) and accrued health-related recurrent expenses (e.g. alcohol, soap, masks), reopening investments/ of the country (1). was an important challenge in Brazil. While it is the responsibility of state and expenses, a category municipal departments to implement actions to ensure the adequacy of school which did not exist Compulsory education: environments, funding and guidelines come from the Federal Government. prior to the pande- Decentralized to the sub-na- mic (non-earmarked Decision-making process within compulsory education levels: The municipa- expense). tional level of first and lities, which in the case of Brazil are a critical and integral part of the institu- second order (i.e. muni- tional architecture for the provision of educational services, also had the au- Criticality of coordi- cipalities). States and mu- tonomy to decide individually about the respective levels of education under nation across levels of nicipalities therefore share their purview (e.g. pre-primary, primary, and lower secondary). Half of the sta- government responsibility for providing tes, for example, called off the school year 2020 much earlier than the original compulsory education. official schedule for that year and planned for a much earlier start of school year 2021. But they could also decide on closures for specific levels of educa- tion within their purview, leading to the possibility that, within the same dis- trict, the municipality decided to close pre-primary institutions, but the state decided to keep partially open upper secondary education institutions (poten- tially located within the same building as a pre-primary counterpart). Mexico Federal Organization: 32 Challenges: Lessons learned: sub-national districts, com- prising 31 states and one Despite the autonomy of states in the administration of their educational sys- Put monitoring mecha- federal district - the Federal tems, the school re-opening process was determined centrally through a “tra- nisms in place for ti- District of Mexico, the capi- ffic light system” for each state. This system was driven by COVID-19 trans- mely decision-making. tal of the country (1). mission rates and managed nationally by the National Health authority. The Advocacy for the reope- reopening situation constantly changed as states changed their epidemiologi- Compulsory education: cal traffic light frequently. ning of schools and return to face-to-face Decentralized to sub-natio- Notwithstanding the lack of “federal organization”-driven bottlenecks, Mexico learning. nal level of first order. faced two key obstacles: • No follow-up and monitoring mechanism to measure participation and learning through distance learning. • Low public awareness on the risks associated with the prolonged closure of schools. Sources: Argentina (OECD, 2016a; UNICEF, 2022a(2)); Brazil (OECD, 2021, UNICEF, 2022a(2)); Mexico (OECD, 2016b; UNICEF, 2022a(2)). (1) Each of the sub-national entities of the three countries can dictate its own Constitution and have the three-tier division of power (Executive, Judicial and Legislative). (2) Updates 31 and 32. Recovering and accelerating: Returning to schooling 96 Two Years After | Saving a Generation Box 4.3. Consolidating more robust hybrid models: An opportunity that cannot be missed Hybrid learning systems should remain not only during the pandemic but also afterwards, in non-emergency con- texts. They have the potential to transform the delivery of education generating flexibility, extending access to basic education, supporting more innovative pedagogical approaches, providing teachers with more training and devel- opment opportunities, allowing longer instructional time, and reducing the cost of learning, among other benefits. Solid investments, planning, participatory approaches, and involvement of all concerned stakeholders are essential to ensure all the population can benefit from flexible models of learning. Hybrid models (where teaching sessions may have some students in class and some participating remotely) should be inclusive, adaptable, designed around different learning objectives, preferences, and realities, and incorporate various pedagogical approaches and tech- nologies. High-quality hybrid models require: 1. Facilitating the interpretation of digital technology in education and training systems through the assessment of digital maturity, assessing skills through digital skills frameworks, and investing in human capital using digital skills policies and programs. 2. Overcoming the digital divides in rural areas and identify hybrid learning models appropriate to every context, including through self-sustained funding models for connectivity. 3. Integration in a national digital strategy which includes reskilling, upskilling, and capacity building to leverage digital technology for life; developing appropriate governance and regulatory frameworks; and creating and pro- moting public-private partnerships. Source: Broadband Commission for Sustainable Development, International Telecommunication Union, and UNESCO (2021). some of these regional experiences and/or implementing Five top interventions to spur enrollment and atten- other policies that can assure that all children are re-en- dance (information; cash transfers; services and learn- rolled and continue attending school thereafter. ing materials): Evidence collected during the pandemic, both from • Conducting inclusive information campaigns on international and regional experiences, shows some school re-opening and re-enrollment, especially for successful policies to spur enrollment and attendance, vulnerable populations, including contacting the and deter dropouts, that are worth noting. Below are student’s families, parents, or guardians to facili- summarized the top 5 interventions (presented in Figure tate and encourage re-enrollment. Several countries 4.5) that were proven to positively impact enrollment and have rolled out back-to-school campaigns to get chil- attendance. It is important to highlight that these inter- dren back to school and learning163. The World Bank fi- ventions build on different incentives, ranging from the nanced these types of campaigns in countries such as use of information, to providing additional services and Madagascar, Ethiopia, and Pakistan, always placing an materials, and cash transfer programs, and these different emphasis on students with the highest risk of dropping interventions have different cost implications (informa- out. In Colombia, a campaign launched in 2021 named tion campaign on one extreme, school feeding and cash “Este Cuento es Tuyo” (This Tale is Yours) tried to in- transfer programs on the other extreme). This implies that centivize children to go back to school, increasing ac- countries should make an informed decision based on cess to and permanence in the education system. This their own diagnostics and funding availability. campaign complemented an already existing program “Todos al Cole” (Everyone to School) which benefitted 163 Ibid. Recovering and accelerating: Returning to schooling Two Years After | Saving a Generation 97 Figure 4.5. Successful policies to spur enrollment and attendance, and deter dropouts Campaigns on school Deploying cash Guarantying that Ramping up school Targeting support for re-opening and transfer programs, learning materials, feeding programs speci c groups re-enrollment especially for information, and disproportionately vulnerable groups services are accessible a ected by the to all pandemic Early Warning Systems Identifying and targeting more than 54,000 primary education children follow- Figure 4.6. Countries in LAC with previous ing their enrollment, through the provision of psycho- experiences with cash transfers logical assistance and extracurricular activities for Cash Transfers those students most likely to drop out164. Drop-outs would likely have been higher without these policies. Argentina Dominican Republic Mexico Bolivia Ecuador Nicaragua • Deploying cash transfer programs, especially for Brazil El Salvador Panama vulnerable groups. These types of strategies, which tie Chile Guatemala Paraguay cash support to families to school enrollment and atten- Colombia Honduras Peru dance, have proven to be effective in different countries Costa Rica Jamaica Trinidad and Tobago of the region such as Brazil, Colombia, and Mexico165. Hard conditionalities could be replaced by softer ones Uruguay to broaden participation. In Brazil, under the program Source: Own elaboration based on Stampini and Tornarolli (2012). Bolsa Familia, cash transfers have been provided to vul- nerable students which, according to the literature and a recent analysis, have led to increased enrollment166. • Guarantying that learning materials, information, and In Uruguay, for example, an inclusive back-to-school services are accessible to all, but especially to people campaign was deployed and entailed cash transfers with disabilities and students from ethnic minority targeted to students at high risk of dropping out of groups. This includes the distribution of books, uni- school. In Jamaica, a similar strategy was developed, forms, and other school supplies. Argentina invested targeting households with children with disabilities, US$5.6 million in 2021 into the National Reading Plan and providing financial support through cash transfers. which distributed more than 1.5 million books to schools These cash transfers programs entail a significant in- in pre-primary, primary, and secondary levels. A total of centive for enrollment and attendance, especially of approximately US$30 million was invested in books and 15-to-17 years old, as they operate as a significant in- other learning supplies under the program “Acompañar: come support to deal with the direct and indirect costs Puentes de igualdad (Teaming: Bridges of Equality)” im- of schooling and have been or have the potential to be plemented to re-enroll students who interrupted learn- applied at scale (See Figure 4.6 for a full list of LAC ing during the pandemic. Improving learning conditions countries for which Conditional Cash Transfers (CCT) at home (including access to devices, connectivity, and programs were implemented to date).167 availability of books) was also highly instrumental to 164 See Colombia’s National Education Ministry (2021) and Páez and Castillo (2021). 165 Attanasio et al. 2021; WBG et al. (2021). 166 WBG (2021c); WBG (2022a). 167 More broadly, CCTs are a significant intervention for poverty alleviation. The Inter-American Development Bank analyzed 19 LAC countries for which CCT pro- grams were implemented in the past and estimated that the poverty headcount index would be on average 13% higher (with variations ranging from 1% in Paraguay to 59% in Uruguay) had these programs not been implemented. See Stampini and Tornarolli (2012). Recovering and accelerating: Returning to schooling 98 Two Years After | Saving a Generation maintain children and youth engaged in the schooling work to supplement family incomes or be the predomi- process. The World Bank’s Read@Home program aims nant gender to support with domestic chores or earlier to deliver reading, learning, and play materials to hard- siblings’ childcaring172. At the regional level, other ini- to-reach homes. In the Dominican Republic, notebooks tiatives that targeted girls were the creation of mate- and books were provided to students enrolled in public rials with a gender lens, which can also help with the schools. According to the survey implemented by that increasing learning gaps. For example, in Honduras, the country’s Ministry of Education, around 89% of the par- Government tried to deploy educational resources with ents and caregivers with their kids in primary, and 81% a gender approach through their online learning plat- with their kids in secondary, reported that their children form, where students had access to different activities used those books and notebooks during the pandem- with a gender focus through a “learning passport”173. ic168. Uruguay invested in social inclusion and equity in education technology through “Plan Ceibal”, distributing To make sure to identify and track students most at risk laptops and no-cost Internet to students and teachers, of not re-enrolling and/or dropping out, several coun- along with digital skills training. tries of the region have also been developing and im- plementing Early Warning Systems (EWS). EWS, although • Ramping up school feeding programs. School clo- very recent and therefore with limited evidence of their im- sures not only interrupted learning but also social pol- pact, have proven to be useful tools in the LAC region to icies such as school feeding programs that seek to pro- date174. They all use individual and family risk factors to vide balanced nutrition to vulnerable children. Before identify students at risk and have helped take timely ac- the pandemic, approximately 85 million children in the tions, implementing more effective strategies to support LAC region used to receive meals from school feeding those students, and undertaking a close follow-up, eventu- programs. For 11% of these children, this meal repre- ally preventing them from dropping out. By doing this they sented their main daily source of food169. During the also support a more efficient use of resources across all in- pandemic, many countries created ways to keep pro- terventions. Previous experiences in the United States and viding students with their meals such as Mexico where Guatemala have shown that this type of systems and inter- during school closures 17 out of the 32 states continued ventions reduced the percentage of students dropping out providing food services with meals to be prepared at from school by between 1 to 4 percentage points175. In Cen- home170. School feeding programs should be prioritized tral America, Belize, Costa Rica, and El Salvador count with as they are an important tool to incentivize the return operational EWS as pilots or at scale. Costa Rica’s EWS uses to school, engage students, and guarantee food securi- an especially wide range of variables related with the fam- ty, especially to the most vulnerable population. ily and social environment to identify youth at risk. Gua- temala and Honduras have developed systems to mitigate • Targeting support for specific groups that have been the risk of drop-out between primary and lower secondary. disproportionately affected by the pandemic, for example, girls. Evidence from the impact of the pan- Chile, Peru, and Brazil have made important strides in demic showed that, for some specific LAC countries, ef- developing EWS. Chile implemented an EWS during 2020. fects on enrollment and attendance disproportionately More than 3,500 educational institutions implemented the affected girls (especially in lower-income households). system and identified 75,000 students at risk of dropping In Paraguay, in 2020, drop-out rates for lower second- out. As shown in Figure 4.7, using administrative data not ary education for girls were 23% higher than for boys171. only about students’ education but also health, and other Studies have shown that cash transfers, among others, factors, the system analyzes each student’s situation through are effective strategies for encouraging girls’ enroll- different dimensions: individual factors, family factors, learn- ment as they help prevent them from resorting to paid ing factors, and community factors. These data are constantly 168 WBG (2022b). 169 IDB (2020). 170 Ibid. 171 Administrative data from the Paraguay’s Ministry of Education show that, in 2020, dropout rates for LSE for girls were 5.3%, an all-time high, compared to 4.3% for boys. For perspective, drop-out rates for boys and girls in 2015 were 4.6% and 3.6%, respectively. 172 Malala Fund (2020). 173 WBG et al. (2021). 174 Perusia and Cardini (2021). 175 Faria et al. (2017); Haimovich et al. (2021). Recovering and accelerating: Returning to schooling Two Years After | Saving a Generation 99 Figure 4.7. Chile: Early Warning System Administrative Data Dimensions Education Individual factors Prioritized identi cation Advanced analitical of students at risk methodologies of dropping out Health Family factors School, learning factors Identi es the top 10% of students at risk of dropping out Community factors Source: Larraín (2021). analyzed using advanced methods and algorithms that are Figure 4.8. Brazil: Early Warning System useful to screen students at risk of dropping out, and then Early Warning System (Brasil na Escola) help take timely actions to keep these students in school. Three steps: Some strategies implemented by schools to prevent drop- outs included home visits, flexibilization of the curriculum, 1. Dropout Risk Questionnaire - Paper-Based (4x per year). provision of textbooks and meals, close follow-up, and long- - 12 yes/no questions (risk factors). term support. Another successful experience is found in Peru - Levels of Risk (number of "Yes"). with “Alerta Escuela” 176 which was launched in October 2020. According to the Peruvian government statistics, 30% of the 2. Qualitative Evaluation institutions’ directors used the platform to revise at-risk situ- - Interviews "high risk" students. ations between March and November 2021. The tool, current- - Identi es qualitative factors. - Follows particular cases. ly being improved for clarity and simplicity, was found effec- tive by 60% of them. Key actions taken upon identification of youth at risk included tutoring, remedial education pro- 3. Menu of Programs grams, and provision of printed materials. As a result, even - Based on risk factors, students are though 428,000 students were found at risk of dropping out, directed to local programs. - "Personalized dropout interventions" it is assessed that 89% continued studying in 2021. Focused on: In 2021, Brazil implemented the Programa Brasil na Es- – Vulnerable schools – Low performing schools cola to support school retention and learning recovery. (>70% Bolsa Familia) (IDEB < 3) The EWS at the core of this program establishes a risk map Source: Adaptation from Sartori (2021). based on multiple factor (to be updated every year), used to then implement targeted complementary interventions (Figure 4.8). The system is expected to be applied to more than 120,000 schools in 2022. A key question for EWS is dropouts are tackled adequately is essential. School to what extent they can be scaled-up to address dropouts systems across LAC must go the extra mile to re-enroll stu- in a more systematic and sustainable way, so it will be im- dents and keep them in school, giving special emphasis to portant to learn from this and other expansions in 2022. age groups more affected during the pandemic, and to the most vulnerable children and youth, encouraging collective A temporary disconnect from school could turn into per- coordination and cooperation between governments, soci- manent drop-out if not treated with a sense of urgency; ety, and media. In the short term, these efforts should focus hence ensuring enrollment, attendance and potential on five core policies: (i) inclusive information campaigns on 176 See https://alertaescuela.minedu.gob.pe/. Recovering and accelerating: Returning to schooling 100 Two Years After | Saving a Generation school re-enrollment, contacting children’s families, par- It is essential that this process is brought to completion in ents, or guardians; (ii) deploying cash transfer programs, countries where reopening is still partial and be sustain- especially for vulnerable groups; (iii) guaranteeing learning able. The school re-opening process needs to be informed materials, information, and services are accessible to all, by the latest evidence of the health situation locally and especially those most vulnerable; (iv) strengthening and be coherent, and clearly owned by principals and all gov- guaranteeing continuity of school feeding programs; and ernment levels. If a full reopening is not immediately pos- (v) targeting support policies for disproportionately affect- sible, earlier grades of education should be prioritized, ed groups of interest (e.g. girls). These policies can be com- and hybrid learning modalities can be a possibility in the plemented by the development and implementation of early short-term. Proper and clearly communicated and funded warning systems to identify and monitor students most at safety standards need to be established for all schools. risk of dropping out, and, understandably and importantly, by policies to recover from learning losses to prevent future Multiple complementary efforts are needed to ensure dropouts related to poor learning. All these interventions re-enrollment of children and youth and keep them in also have the potential to increase attendance and decrease school. Special attention will need to be given to mon- dropouts in the long-term. itoring enrollment during 2022, to the age groups more affected during the pandemic, and to the most vulnerable children and youth, who are the most at risk of dropping out. Policies will have to include inclusive information 4.3 Key takeaway messages campaigns, leverage cash transfer programs, and support access to learning materials, services, and school feed- Completing the school reopening process is the first ing programs, with a special focus on the most vulnera- most important step that must be taken to get children ble groups. Building on the emerging country experienc- and youth back to schooling. Countries in the region have es, early warning systems can help identify, monitor, and been progressively reopening their school systems and target students most at risk of dropping out, also for lon- prioritizing a return to face-to-face classes for students. ger-term decreases in the out of school youth population. Two Years After | Saving a Generation 102 Two Years After | Saving a Generation PHOTO BY: © GONZALO BELL / SHUTTERSTOCK.COM Chapter 5 Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 103 A s reviewed previously, recovery must focus on the essentials: returning to schooling and recovering from learning losses (Figure 4.1). Getting back and stay- Figure 5.1. Three-dimensional model of curriculum, pedagogy, and assessment ing in school is only the first step to recover from the ef- Curriculum fects of the pandemic on education: children also and fore- O cial policies that state the most need to get back to learning. Without learning it is knowledge, skills, competencies, values, and behaviors that also likely that schooling itself will suffer. Recuperating lost students are expected to learn. or forgone learning, and make up for the lost time, will re- quire critical choices and interventions. At the same time, a deterioration of mental health and well-being of children Assessment and youth across education levels has hindered their read- Process of Teachers gathering evidence Students iness to learn and acquire both foundational and transfer- on student learning Pedagogy able skills and could also lead to increased dropouts and and judging the Teaching practices degree to which that shape students’ risky behaviors. The well-being of teachers and parents students are experiences and opportunities have also suffered. And, while the crisis may have been the meeting to learn in order to meet curricular curricular expectations. opportunity to improve digital skills, there is a still a long expectations. way to go to address digital gaps across all education levels for recovery and acceleration. Addressing these challenges also requires prompt and continuous action. Pedagogy shapes what, how and when to assess. Assessment results inform how to adapt teaching to the learning needs of This chapter delves into the “recovering from learning students. The broader social context shapes education aims, policies, losses” agenda, while also briefly discussing actions institutions, and the physical and social conditions in which to address psychosocial and digital gaps. It will discuss teaching and learning occurs. critical actions related to: (i) curricular consolidation; (ii) assessment of students’ current learning levels to prop- erly diagnose the breadth, depth, and characteristics of learning losses; and (iii) building on the curricular deci- information to inform the changes needed in schools and sions and on the evaluation of learning losses, interven- education systems. In most effective school systems, these tions that can help provide the needed (and presumably three pillars are aligned, feedbacking one each other and accelerated) learning recovery. When discussing founda- making synergies. They operate as an integrated system tional skills, the chapter will also relate when relevant to that puts students at the center and provides strong sup- transferable skills, as defined in Chapter 3, insofar as they port to teachers to support students (Figure 5.1). play an important role in the ability to accumulate skills, including foundational ones. The chapter will also provide All three learning pillars where greatly shaken by the initial insights on directions for improvement in the long COVID-19 pandemic. In addition to the disruptions in at- run. Finally, it will also discuss some critical actions to de- tendance and engagement, school closures precluded cov- velop psychosocial health and well-being, and review ac- ering the curriculum, teachers needed to adapt pedagogy tions to address digital divides. to remote education, and assessment results were not avail- able to inform learning and teaching, resulting in the large expected and actual learning losses highlighted in previous chapters. These issues were made worse by pre-existing in- 5.1 Recovering from learning losses stitutional constraints and service delivery failures especial- ly related to weak leadership at the school and system lev- Children go to school to learn. Schools are the place els and lack of incentives to generate and use data at scale where they acquire academic knowledge and skills, where for evidence-based decision-making. Most countries were they learn to care about others, where they learn to learn. pro-active in adopting mitigating measures, but the agen- School systems usually foster learning based on three da for action is still long to put children and adolescents main pillars or dimensions: a curriculum that states what back on a recovery and acceleration path. This section pres- students are entitled to learn; a pedagogy that guides ents the challenges countries faced on each dimension, the how to teach towards curricular goals; and assessments practices they have been putting in place to address, or at that provide evidence on student learning and invaluable least mitigate, learning losses generated by school closures, Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 104 Two Years After | Saving a Generation and how to build on those and pre-pandemic practices to •  Curricular expectations that are not realistic given the speed-up the learning recovery process. While the three real learning level of the students, with most students pillars continuously feed into each other, and it is therefore falling well behind the curricula. important to keep in mind this articulating framework, the sequencing of the discussion adopted in this chapter aligns •  One size-fits all curricula that do not recognize the with the urgent agenda for action around consolidating, as- fact that students do not learn at the same pace, that sessing, and taking remedial interventions, with a focus on teachers have very heterogeneous classes with stu- the short-term, while highlighting potential opportunities dents at very different learning levels. for long-term improvement. •  Teachers and supervisors being more concerned with 5.1.1 Prioritizing the curriculum “covering the curriculum” than with student learning. The pandemic opened the door for countries to make •  External control and accountability mechanisms put- substantial revisions to their curricular policies, prior- ting more incentives on teachers covering the curricu- itizing what was considered essential learning for all lum rather than on ensuring learning for all. students. Before the pandemic, there was criticism that, in general, the curricula of the countries were too broad, Even as it led to substantial learning losses, the pandem- demanding, and rigid. However, there was neither political ic paradoxically paved the way to address some of these nor technical support for narrowing down curricula that curricular concerns. When school closures and remote targeted the development of high-level competencies177. learning became imminent, countries, to the extent possi- With the pandemic, a turning point was reached that made ble, reacted by: (i) identifying “essential learnings” to be the narrowing down of the curriculum a necessity. There taught to all students, with a drastic reduction of the curric- was the urgent need to address school closures, a reduced ular content, skills, competencies to be taught; (ii) adopting school year calendar, and adjustments in pedagogy to a more deliberate focus on foundational mathematics and adapt for remote learning. There was no way the whole language skills; and (iii) allowing for greater flexibility in curriculum could be covered in new digital, television or the implementation of the prioritized curriculum. radio education programs. Several LAC countries decided to simplify their curric- Curricular documents in Latin America are relatively ula, focusing on key subject areas and “essential learn- homogeneous in their disciplinary and pedagogical ap- ings”. According to the last Survey on National Education proaches. More or less, curricula expect students to learn Responses to COVID-19 School Closures180, 8-10 countries similar contents, skills, competencies and attitudes. They made adjustments to their school calendars and curricula emphasize solving real life problems and use a communi- by education cycle, increasing slightly between 2019/20 cative and socio-cultural approach to teach languages. They and 2020/21 (Figure 5.2). Teaching mathematics and lan- endorse an active pedagogy that promotes students build- guage (reading, writing, literature) became a priority over ing their own knowledge based on their experiences, in real other subject areas. Brazil, Ecuador, Honduras, Panama, and significant contexts.178 However, there were concerns and Saint Lucia are among those countries that prioritized about curriculum design and implementation179, including: curricula in all three cycles of the school system (primary, lower secondary and upper secondary) and during the two •  Curricula too ambitious and complicated, covering school years most affected by the pandemic. too many subjects and skills, without a clear focus or sequence. Countries should continue prioritizing their curricu- lum to recover learning. As schools have been reopening, •  Curricula with an excessive focus on content itself to countries should keep focusing on the key subjects of lan- the detriment of foundational and transferable skills. guage and mathematics, as well as on transferable skills 177 Valverde et al. (2020). 178 UNESCO-OREALC (2020). 179 See, for example, GEEAP (2020) and (2022). 180 UNESCO, UNICEF, WBG, and OECD (2021b). The 26 countries that answered this survey were: Antigua and Barbuda, Bahamas, Belize, Brazil, British Virgin Is- lands, Cayman Islands, Colombia, Costa Rica, Cuba, Curaçao, Dominican Republic, Ecuador, El Salvador, Grenada, Guyana, Honduras, Jamaica, Mexico, Nicaragua, Paraguay, St Kitts and Nevis, St Lucia, Suriname, Trinidad and Tobago, Turks & Caicos Islands, and Uruguay. Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 105 Figure 5.2. Number of LAC countries that prioritized Figure 5.3. Programmatic intervention areas through certain areas of the curriculum or certain skills due to multiple pathways COVID-19, by year and school cycle, 2021 12 Systems Strengthening 10 10 10 9 Number of countries 8 8 8 Enable Environments 8 6 Teaching and Learning 4 2 Skills Development for all Children and Adolescents 0 Primary Lower Upper Primary Lower Upper Secondary Secondary Secondary Secondary Source: UNICEF (2021e). 2019/2020 2020/2021 Source: UNESCO, UNICEF, WBG, and OECD (2021b). Survey on National Ed- ucation Responses to COVID-19 School Closures, round 3 (Feb-Apr 2021). a set of complementary interventions focusing on increas- Overall, 26 LAC countries answered the survey. ing readiness to learn, revising the curriculum, enhancing teacher support, and improving management for learning (see Box 5.1). such as critical thinking, as a means to support learning. They should also keep prioritizing teaching foundational Moving forward, the crisis may also the opportunity to skills that tap the actual proficiency levels of their students. reiterate the importance of transferable skills and the For instance, if many third graders cannot still read simple need for a comprehensive vision and approach to de- stories according to curricular expectations, these students velop then. Transferable skills development needs to oc- should have the right learning opportunities to allow them cur across the life-long learning cycle and through multiple progress in their literacy skills. These adaptations should learning pathways. Transferable skills development is a cu- be done even if they depart from curricula and programs mulative process from early childhood through adolescence of study. Adapting curriculum and pedagogy could be fur- to adulthood, that can occur through different modalities ther reinforced by increasing teaching time, especially for including formal, non-formal and informal education. To students with greater learning losses. As further discussed improve the development of transferable skills, UNICEF´s below, two interesting examples of reforms along these Global Framework on Transferable Skills proposes three lines are found in the learning recovery program “Acelera intervention areas for action: (1) Teaching and learning, Brasil”181 and in the State of Ceará in the same country182. including appropriate and context aligned curriculum, con- tent, pedagogical practices and assessments183 - an exam- Moving forward ple from Argentina can illustrate pedagogical approaches to develop transferable skills (Box 5.2); (2) Enabling envi- Refocusing on foundational skills should also be a pri- ronments, where students feel physically, socially and emo- ority for the long-term. This type of reforms is critical tionally secure to participate and can develop skills through considering that, even before the pandemic, more than positive relations with peers and adults; and (3) System half of students did not meet minimum proficiency lev- strengthening, including strategies, plans and policies at els according to ERCE 2019. There is an opportunity to the national and local level, budgets and human resources, build back better to enhance foundational skills through among other (Figure 5.3). 181 Mancebo and Vaillant (2022). 182 Loureiro, Cruz et al. (2020), Loureiro, di Gropello et al. (2020). 183 One example of intervention that needs focus is teacher training. Even though over the last decade socio-emotional skills have been increasingly getting more attention across the region’s education systems, a pre-pandemic study of 12 LAC countries found that pre-service and in-service teacher training programs offered scarce opportunities in transferable skills. Arias at al. (2020). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 106 Two Years After | Saving a Generation Box 5.1. Accelerating: Foundational skills for the future The lack of foundational skills among LAC students, particularly in connection with low reading comprehen- sion levels, was a key limitation for remote learning during the COVID-19 pandemic. This was particularly true for the most disadvantaged students, who did not have access to high-tech online platforms and synchronic virtual education. To overcome both preexisting and post-pandemic challenges, and significantly reduce “learning poverty” in the region, education systems must focus, effectively, on developing basic levels of proficiency in reading, writing, and mathematics among all students. A set of key policy actions to enhance learning opportunities in the region is presented below, grouped in four main areas: 1. Increase readiness to learn: The acquisition of foundational skills in reading, writing, and math begins before formal schooling. A first necessary step to improve foundational skills is to increase quality in early childhood development and in preschool interventions with special emphasis on the most disadvantaged children and at-risk population. Interventions such as home visits and preschool are cost-effective and their effectiveness can be increased over time, through sustained improvements in teacher and classroom quality184, and direct instruction185. Making these interventions easily accessible and affordable to all parents, particularly women, is also key for their overall success186. 2. Improve teacher quality and effectiveness: Evidence from the region has shown that teacher quality is of grave concern187. Recent studies in five Central American countries have shown that teachers are not receiv- ing adequate initial teacher training. It is imperative to provide teachers and school administrators with a systematic set of tools that will allow them to foster children’s learning so they can reach their full potential. Teachers who have not received quality initial teacher training will most likely lack the tools to adequately assess what skills their students are bringing to their classrooms and which ones they are missing. Learning difficulties, however, can be prevented in young, at-risk students when timely assessments and interventions are deployed. Hence, giving teachers the tools they need to assess student learning and its progress in an effective and objective manner is key. Other recommendations are structured, and scripted lesson plans which have been proven to be effective tools to increase learning, giving teachers and children textbooks and educational resources to improve learning. 3. Revise the current curriculum and instructional time to emphasize critical literacy (reading and writing) skills: Reading, writing, speaking, and understanding texts are fundamental building blocks for learning that need to be properly developed. • For alphabetic languages, the Science of Reading (SoR) has shown that effective reading programs and instruc- tion are key enablers for strong literacy skills acquisition for children, from preschool to the first grades of ele- mentary school. SoR has debunked many of the methods that, over the years, were and are still used in reading instruction. Revising methods like that of whole-language approach, and changing them, although not an easy task, is fundamental to eradicate learning poverty in the long term. (Box continues on next page) 184 Araujo et al. (2016); Nores et al. (2018). 185 Näslund-Hadley et al. (2020). 186 Mateo-Díaz and Rodriguez-Chamussy (2016). 187 Elacqua et al. (2018); Burns and Luque (2014); García et al. (2014). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 107 Box 5.1. Accelerating: Foundational skills for the future (continued) • Moreover, LAC children also suffer from “writing poverty”. Very limited time is devoted to writing per se, and writ- ing instruction, in the classroom. Therefore, effective reading and writing instruction require sufficient time to be allocated to these activities, both daily and throughout the academic year. In the case of reading, an estimated 600 hours of instruction over a three-year period are required to acquire mastery188. This needs to be reflected in the curriculum and in how teachers, schools, and district administrators organize instructional time, particularly in the first three years of primary schooling. • Last but not least, schools should foster active mastery of speaking and presentational skills. Being able to artic- ulate clear ideas and present them orally to an audience, in a short period of time and in a focused manner, is an essential competence today. Teaching and training students on these abilities must be a critical ingredient of the foundational skills development process. 4. Refocus system and school management towards foundational skills development: Literacy, numeracy and transferable skills such as problem-solving skills are the cornerstone of learning. Hence, management for learn- ing – and for developing these competencies - is fundamental. Figure B.5.1.1 presents some key aspects that may guide the refocusing of school management on learning such as: (i) providing technical support to the school leadership team; (ii) implementing individualized monitoring and formative assessment systems; (iii) provid- ing basic resources for effective teaching and instruction programs; and (iv) allowing for some flexibility at the school and district level so that focus can be on learning. Supporting school principals and better aligning and allocating roles at all levels of the education system articulating those around learning is also at the core of a recent regional study on management for learning189. Figure B.5.1.1. Key actions for refocusing school management on learning outcomes and foundational skills Individualized monitoring Providing technical Allowing for exibility and formative assessment Providing basic support to school at the school and system, that allows to resources so that principals and their district level, so that all evaluate student’s e ective teaching and teams to enhance the focus can be on learning progress instruction can occur. learning learning throughout the year Reducing student/teach- Should be structured Learning outcomes er ratios in the earlier based on school should be reported to Textbooks, books, school grades, while increasing characteristics, such the educational supplies and additional them at the higher ones, as size, location community at large and personal. redistributing students (urban, rural), etc. decision makers. based on their skill level, and not by age or grade. The Response to Intervention (RtI) model and the Teaching at The Right Level (TaRL) approach are clear examples of mechanisms through which district-level and school-level management can be refocused on learning outcomes and foundational skills. RtI is a tiered intervention model that aims to ensure that all students receive the support they need to succeed academically. Over the last decades, school districts in Canada and the United States have implemented RtI initiatives with promising results190. Pratham, an Indian NGO, developed the TaRL model to target learning poverty at the primary school level. The model is aimed at improving foundational skills, regardless of age or grade level. The key elements of this approach are presented on Figure B.5.1.2. (Box continues on next page) 188 Crawford and Oviedo (2022). 189 Adelman and Lemos (2021). 190 Connor, Morrison et. al. (2013), Connor, Phillips et al. (2018). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 108 Two Years After | Saving a Generation Box 5.1. Accelerating: Foundational skills for the future (continued) Figure B.5.1.2. Teaching at the right level (TaRL) KEY ELEMENTS OF THE APPROACH 1. Simple one-on-one 3. For each group there are a assessment done to group set of activities and Std. III, IV and V children by materials appropriate for level rather than by grade. their level. 4. Similar assessment used for Children learn in groups and tracking children’s progress, also individually. Teachers or periodically and to guide instructors do activities with 2. Children’s groups are onsite monitoring. groups. made according to the basic assessment. Available teachers or instructors allocated to As children make progress facilitate group activities they move into the next and to guide children’s group. work. Source: https://www.pratham.org/about/teaching-at-the-right-level/ Sources : Adelman and Lemos (2021); Appelbaum (2009); Araujo et al. (2016); Banerji and Chavan (2016); Burns and Luque (2014); Connor, Morrison et. al. (2013); Connor, Phillips et al. (2018); Elacqua et al. (2018); García et al. (2014); Mateo-Díaz and Rodriguez-Chamussy (2016).; Näslund-Hadley et al. (2020); Nores et al. (2018); Chakera et al. (2020); Fuchs and Fuchs (2006); Jiménez et al (2021); Loureiro, Cruz et al. (2020), (2020b); Mourshed et al. (2010); O’Connor et al. (2012); Ontario Ministry of Education (2013); Rodrigues Da Cruz Boari and Crawford (2022). Box 5.2. PLaNEA’s Program: An innovative approach to develop transferable skills Aiming to promote learning and skills development in lower secondary education in Argentina, UNICEF Argentina has developed the PLaNEA program. PLaNEA focuses on supporting the adoption of an innovative model of inquiry-based learning for adolescents in secondary schools in urban areas. The strategy changed teaching and assessment practic- es through the introduction of project-based pedagogy accompanied by formative evaluations in Social Sciences, Nat- ural Sciences, Mathematics, Spanish, and Computer Science. Its implementation involved in-service teacher training, the establishment of peer groups for both teachers and students, and the development of a component on adolescent health care. Moreover, to ensure the educational leadership necessary to sustain an innovative approach, school net- works comprised of school directors, supervisors, and school counselors were formed to facilitate exchange on com- mon issues of concern. Other important components included the provision of technical assistance to public officials to adjust regulations such as those related to school schedules, learning environments, teacher work, and promotion criteria. PLaNEA also seeks to use data (academic grades, attendance, surveys) to identify issues that require special attention and ensure policies and interventions are evidence-based and specific. The strategy has shown positive results in education. This initiative began in 16 schools in 2018 and, as of 2022, it already covers 193 secondary schools in three provinces (Tucumán, Chaco, and Córdoba), reaching over 40,000 students, more than 2,400 teachers, and 320 directors. The main results to date include significant improvements in the school trajectory indicators of 1st year students compared to the average of schools in each of the provinces. Compared to the average of state secondary schools, schools under the program reduced school dropout rates by half, increased school promotion by 12 percentage points and reduced repetition by 17 percentage points. Internal surveys show that the vast majority of teachers surveyed (more than 80%) indicate that they observe greater participation of students in classes, greater enthusiasm, and commitment to activities. The initiative is currently under evaluation. Sources: UNICEF (2019b); UNICEF (2021c). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 109 5.1.2 Assessing learning levels and losses Bank, UNESCO, and UNICEF has been working towards im- plementing a new version of ERCE post-pandemic (2022- COVID-19 faced LAC countries with unprecedented 2023) with interested countries to analyze the impact of challenges to maintain student assessment191 practices the pandemic on different population groups to better tar- at a juncture where they were more than never need- get education policies. ed. Classroom assessments could not be continued if stu- dents were no longer in the classrooms. School diplomas Global international assessments include an agenda and examinations became unmanageable with school clo- of innovations that was set well before the pandemic. sures. Some national and international assessments had PIRLS, PISA, TIMSS are all evolving towards digital as- to be postponed, modified, or ultimately cancelled. At the sessments that allow for more interactive real-life ques- same time, monitoring and supporting learning was more tions and problems, and for assessments that are more urgent than ever. After months of school closures and re- fitted to the skills of the students (e.g., adaptative tests). mote learning, damage has been done. Data was and is PISA 2022 will explore the measurement of creative think- needed to diagnose the magnitude of learning losses and ing. ICCS will take a deeper look into climate change, pop- to inform the decision making for adequate actions. Be- ulism, and authoritarianism. low are reviewed the delays, changes, and innovations adopted by LAC countries during the pandemic for differ- National assessments for ent types of assessment, and recommendations to move monitoring purposes forward. Overall, learning assessments should be urgently re-prioritized, with a focus on formative assessments and In LAC, national assessments for monitoring purposes complementary standardized international, regional and/ usually refer to large-scale programs to measure learn- or country assessments. Each assessment type has its own ing progression towards achieving curricular objec- characteristics and strengths which countries should care- tives193. Standardized tests and questionnaires are usual- fully assess when making a decision of what and how to ly externally administered every 1-5 years to all students introduce and scale-up. (census) or to nationally representative samples of stu- dents, using paper and pencil or computer-based format. International assessments for Results are reported at the country, regional and/or school comparing education systems levels. More recent innovations have included reporting learning levels describing what students can do at each As seen previously, the calendar of administration of level; and learning trends over time. Results can be used international assessments192 was kept mostly on track, to inform educational policies, school planning and, to a but it will take time to have results to guide remedi- lesser extent, pedagogical practices. In some countries, al actions. PISA was postponed from 2021 to 2022 in 14 results are “low-stakes” in the sense that no strong con- LAC countries. However, the administration of the other sequences are associated to them (e.g., ARISTAS in Uru- international assessments was not altered. TIMSS (Trends guay). In others, results are “high stakes” given that they in International Mathematics and Science Study) was ad- are used for accountability purposes, i.e., allowing or not ministered in Chile in 2018, before COVID-19, and will be schools to operate, or distributing economic or symbolic administered again in 2023. PIRLS, despite the difficul- incentives such as awards, and public recognition (e.g., ties, was administered as planned in 2021 in Brazil and the Quality Assessment System – SIMCE, in Chile). Chile and is scheduled to be administered again in 2026. The regional assessment ERCE was administered in 16 Before the pandemic hit, almost all LAC countries al- LAC countries in 2019 (before the pandemic). It will be ready had in place a national assessment for monitor- implemented again 2025 for all countries participating in ing learning. These usually reported important learning LLECE. Moreover, the learning alliance between the World gaps between what students actually knew and could do 191 In the school context, student assessment refers to the process of gathering evidence about student outcomes in order to monitor, certify, select and/or make pedagogical decisions. Student outcomes have been traditionally understood as the knowledge, skills, and/or competencies that students acquire in different aca- demic subjects (e.g., language, mathematics, sciences). 192 International assessments refer to global (e.g., PISA, PIRLS) and regional (e.g., ERCE) programs to monitor learning and to make comparisons among countries and/or education systems. They administer tests on subject areas (e.g., mathematics, language, science) focusing on the content, skills, and competencies most valued by countries. Tests and questionnaires are administered to nationally representative samples of students. Results are reported at the country level and are used for monitoring purposes only (low-stakes). Many LAC countries participate regularly in one or more international assessments. 193 Ramírez and Valverde (2021), Valverde and Ramírez (2021). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 110 Two Years After | Saving a Generation vis-à-vis curricular expectations. They also highlighted Figure 5.4. Map of countries based on decisions made inequalities along socioeconomic lines, ethnicity, and ur- regarding the administration of learning assessments ban/rural status. These assessments were critical to raise in 2020 and 2021 alarms about the learning crisis even before the pandemic. When the pandemic hit, several LAC countries reacted by cancelling or reducing their national assessments for monitoring learning, as well as other large-scale assessments, including examinations to certify pri- mary or secondary education, and their university en- trance examinations. A study commissioned for this re- port of how countries reacted to the pandemic194 shows that only Brazil, El Salvador and Uruguay administered all their large-scale assessments as planned in 2020. As shown in Figure 5.4, some countries implemented at least one assessment (Chile, Colombia, Ecuador and Guyana), and others none (Argentina, Belize, Costa Rica, Domini- Did implement all Did not implement any can Republic, Guatemala, Honduras, Panama and Peru). the corresponding of the corresponding learning assessments learning assessment Uruguay was the only LAC country that administered a national assessment for monitoring purposes in 2020 Did implement at least Not applicable one of the corresponding (computer based “ARISTAS”); all the other large-scale as- learning assessments sessments administered that year were secondary school Source: Herrero-Tejada et al. (2022). examinations or examinations for selection into university Note: The map reports decisions made regarding: (a) national assessments or tertiary education. While recognizing the effort of some for monitoring purposes, (b) Examinations to certify and select individual students into primary, secondary and/or tertiary education. countries to at least keep the assessments in some format (see Box 3.2 on Colombia/SABER in chapter 3), this was a drawback in countries’ capacity for monitoring learning. The paradox is that countries were more willing to drop the lack of results yet for most 2021 assessments is a key their national assessments for monitoring learning rather constraint to take immediate recovery actions. In 2022, all than their certification examinations. but two countries (Belize and Ecuador) were expecting to administer all their assessments as planned, but the situ- However, the recovery of national assessments was ation may still evolve196. It will be critical to maintain the fast, and it will be critical to keep this on track. By 2021, 2021 and 2022 assessments on track. most of the countries were back on track administering all their large-scale assessments as planned, including na- There were adjustments in some assessments follow- tional assessments for monitoring purposes, secondary ing the pandemic. As a reaction to the pandemic, several school examinations and university entrance examina- countries adjusted their large-scale assessment programs, tions. Chile and Guyana reported administering at least including national assessments for monitoring purposes, one large-scale assessment, while Paraguay, Peru, Domin- and examinations to certify and select individual students, ican Republic, and Belize opted for not implementing any in terms of assessed areas, modalities, formats and addi- of these assessments in 2021.195 Most notably, 2021 was tional information (see Figure 5.5 for a few of these chang- marked by the return of national assessments for moni- es). The administration of student questionnaires focused toring purposes in several countries, including Argentina, on the general well-being of students received extra atten- Honduras, and Panama. These are positive news although tion during the lockdowns. 194 Herrero-Tejada et al. (2022). Overall, 20 countries were included in this study: Argentina, Belize, Bolivia, Brazil, Chile, Colombia, Costa Rica, Cuba, Ecuador, El Salvador, Guatemala, Guyana, Honduras, Mexico, Panama, Paraguay, Peru, Dominican Republic, Uruguay and Venezuela. 195 Ibid. 196 It is worth noting that Chile’s Ministry of Education has recently asked the Education Council to cancel SIMCE 2022, but the Education Council has not yet made yet a decision. Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 111 Figure 5.5. Continuity, changes, and innovations as a reaction to the pandemic in national assessments for monitoring purposes and examinations to certify and select individual students Continuity Changes and innovations Assessed areas and recipients Mixed or digital format Only modi ed in Saber 3,5,9 (Colombia) where environmen- The format is changed in Enem (Brazil), Saber 3,5,9 tal education was introduced (not related to the pandemic) (Colombia), Graduados (Guatemala), Avanzo and Aprender (Argentina) where in 2021 changes were going (ex PAES, El Salvador), and Transformar (Ecuador) to be implemented for secondary level, but were nally implemented in primary In-person modality Complementary information about teaching and Only modi ed in Ecuador, where the already implemented learning during the pandemic digital format started to be implemented from home Aprender (Argentina), Enem (Brasil), Aristas (Uruguay) Source: Herrero-Tejada et al. (2022). Examinations to certify and examinations. In primary education, 8 countries reported select individual students cancelling the examinations and using an alternative ap- proach for high-stakes decision making (e.g., calculated There are several countries in LAC where students have grades), compared to only 3 countries in upper secondary. to pass a certification examination197 in order to grad- These changes were mainly implemented in Caribbean uate from primary and/or secondary education, and countries with a strong examination tradition. to move into the next education level. Upper second- ary school examinations may also be used for selection During the pandemic, LAC countries with certification into university or tertiary education. In other countries, examinations were confronted with the tough decision students have to take a separate examination to apply to of administering or cancelling these assessments. On university or tertiary education. Examinations are a key the one side, it was reasonable to cancel them given that component of the education systems of these countries. students had fewer opportunities to learn and prepare for They greatly shape what is taught and how it is taught in the examinations, and social distancing and health con- classes. They are high-stake assessments needed to tran- cerns made them hard to administer. On the other side, sition into secondary and/or tertiary education and to ob- in some countries these tests are so engrained in the na- tain good quality jobs. tional culture that it was unthinkable not to have them. Moreover, examination results are tight to selection de- The last Survey on National Education Responses to cisions into secondary and tertiary education. Without COVID-19198 shows that LAC countries made several examinations, countries would have been forced to put changes to their certification examinations due to the in place alternate selection mechanisms (e.g., calculated pandemic, especially in upper secondary (Figure 5.6). grades). Doing so in a rush was risky both technically and This included the introduction of additional safety mea- politically. sures such as extra space between desks; adjusting the mode of administration, e.g., computer-based; postponing Beyond the pandemic, there is concern about the bene- or rescheduling examinations; introducing alternative as- fits of primary and secondary certification examinations sessments, e.g., portfolio; and adjusting the content of the in LAC, given that many countries are still struggling 197 A few countries have primary school certification examinations (e.g., Belize, Guyana, Haiti). Lower and/or upper secondary school examinations are not common in South America but are used in several Central American and Caribbean countries, such as Costa Rica, Dominican Republic, and Jamaica. In the En- glish-speaking Caribbean countries, upper secondary examinations have a double function of certifying end-of-school and selection into tertiary education (see ex- aminations from the CXC, which include 16 countries: Anguilla, Antigua and Barbuda, Barbados, Belize, British Virgin Islands, Cayman Islands, Dominica, Grenada, Guyana, Jamaica, Montserrat, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, Trinidad and Tobago, and Turks and Caicos Islands). In other countries, students take national university entrance examinations after graduating from secondary school (e.g., Brazil, Ecuador). 198 UNESCO, UNICEF, WBG, and OECD (2021b). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 112 Two Years After | Saving a Generation Figure 5.6. Number of LAC countries that adjusted certification examinations due to the pandemic during school year 2019/20 and 2020/20, by type of adjustment 18 17 16 15 14 12 Number of Countries 12 11 11 11 10 10 10 9 9 8 8 8 8 7 6 5 4 4 3 3 3 3 2 1 0 Postponed/ Adjusted Adjusted Introduced health Introduced Canceled exams Other rescheduled exams administration and safety measures alternative and used alternative exams (e.g., content, mode (e.g., (e.g., extra space assessments approach (e.g., extent) computer, between desks) (e.g. portfolio) calculated grades) online) Primary Lower Secondary Upper Secondary Source: UNESCO, UNICEF, WBG, and OECD (2021b). Survey on National Education Responses to COVID-19 School Closures, round 3 (Feb-Apr 2021). Overall, 26 LAC countries answered the survey. to provide education for all. A dramatic case is offered the main source of evidence for teachers to measure ac- by Haiti, a country shaken by political, economic, and so- tual learning and learning losses200. When schools closed, cial crises. Here, only approximately 60% of children fin- classroom assessment practices were disrupted. Teachers ish primary education (école fondamentale) and only 10% could no longer ask their students questions in classes, finish secondary school (baccalauréat). Students have to observe if they could read a story with fluency and com- pass an examination in order to get their diploma and prehension, or administer a quiz to check if they under- move on to the next education level. Around 10% of the stood a concept. Teachers could not comply with school students do not pass the examination at grade 9 (exam- requirements to grade their students in order to inform en de fin école fondamentale), and many end up dropping decisions about grade promotion or repetition. They could out from school. In secondary school, failing rates for the not check on the physical or mental well-being of their stu- baccalauréat are around 25%. In 2021, the examinations dents neither. were postponed, and social distancing measures were ad- opted due to COVID-19. Examinations conditions suffered. Some of these disruptions were overcome with inno- Some administration centers had to be relocated due to vative approaches to classroom assessments. Teachers gang violence. sent weekly materials to their students via email, or talked with their students online, on WhatsApp or on the phone. Classroom assessments Digital education allowed teachers to monitor student per- conducted by teachers formance online, and also to have a grasp of their physical and mental well-being at home. These are in fact among Classroom assessment is by far the most important type the many innovative strategies Chilean teachers adopted of assessment, and it was also the one most severely as an option to more traditional classroom assessment affected by the pandemic199. Classroom assessments are techniques201. 199 UNESCO-OREALC and UNICEF-LACRO (2022). 200 It is also the main source of evidence for teachers to make instructional decisions and to grade students. These assessments can be formative (e.g., when used to give feedback to students) or summative (e.g., when used to grade students at the end of unit), formal (e.g., an end of unit test) or informal (e.g., when a teacher asks a question and students raise their hands); and can be used as diagnostic, intermediate or final assessments. 201 MINEDUC-Chile(2020). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 113 With schools reopening, teachers should be encour- monitoring learning. Students could take them on paper aged to perform diagnostic classroom assessments, and or in digital format, online or offline, at school or at home, should be offered a variety of resources and training to at the beginning or end of the school year. It was up to do so. School leaders and teachers should prioritize diag- the schools to decide. During the school closures, teach- nosing the actual learning level of their students in order ers could drop physical copies of the assessments at the to target instruction202. Teachers should be offered a menu homes of those students who could not access them on- of options for them to explore, select and adapt to their line. Teachers then collected and scored the tests and up- own context. For instance, item banks, assessment proto- loaded the data on an online platform. These assessments cols (e.g., for measuring reading fluency) and teaching and have also strong potential to measure learning losses and learning guides, accompanied by workshops or webinars on help with the recovery effort as schools are reopening. how to plan and implement assessments to support learn- ing recovery. These resources could be homemade and/or There has been a growing interest to offer centrally de- compiled from publicly available repositories203. Toolkits signed formative assessments to continuously monitor should encompass assessments that measure a wide range learning during the pandemic and start recovering. El of skills, including basic literacy and numeracy skills. Salvador, the Dominican Republic, the Province of Men- doza in Argentina, and the state of Guanajuato in Mexico In the short-term, it would also be important for teach- are all taking their first steps along these lines. Table 5.1 ers to complement the learning evidence collected lists some longer and shorter-term formative assessments through classroom assessment with other sources of adopted in LAC countries. Useful insights can be derived evidence, like centrally designed formative assess- from a review of these programs. ments (see next section). Both together could be used to better diagnose the real learning levels of students, Formative assessment initiatives in LAC differ from including recent learning losses, to adjust pedagogy to one education system to the other, but important com- the diverse learning needs of the whole class, and to the monalities can be identified. These assessments priori- unique learning needs of each student. tize measuring foundational numeracy and literacy skills along the whole school trajectory (from grades 1-2). They Centrally designed formative assessments are available in different test formats (i.e., paper and pen- for internal use of the school cil, digital online and offline), and may integrate more traditional items or problems (e.g., using multiple-choice The pandemic accelerated an already existing trend questions) with more innovative ones (e.g., individual toward offering centrally designed formative assess- assessments protocol for measuring reading fluency).204 ments to schools for their own internal uses. In the last They all involve teachers during the assessment cycle. decade, countries like Brazil, Chile, Mexico and Uruguay Participation is a way to empower and train teachers on introduced assessments that are made available for the how to operationalize the curriculum and measure learn- schools and teachers to administer, score, analyze, and ing indicators, on how to diagnose the learning levels of use to improve pedagogical practices. These assessments their students, and how to use results to target pedagogy are meant to complement already existing ones (e.g., to the learning needs of their students. This is in fact the classroom assessments, national assessments, examina- approach adopted by the Dominican Republic in design- tions for certification), and are not meant to grade stu- ing its formative assessment program. Box 5.3 provides dents or to evaluate teacher performance. They also over- further details for the assessment of the State of Ceará in come important weaknesses of national and international Brazil which also illustrates some of these aspects. assessments. Formative assessments also have the potential to as- In the education systems that already had centrally sess some critical transferable skills. Both Chile and designed formative assessments before the pandemic, Colombia have included socio-emotional skills in their these offered a viable alternative for schools to keep national formative assessment initiatives implemented 202 Luna-Bazaldua et al. (2020). 203 Free and publicly available learning assessment resources can be found in the Inter-Agency Network for Education in Emergencies, Learning Pack 7; the FLN Hub, and many others. 204 Herrero-Tejada et al. (2022). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 114 Two Years After | Saving a Generation Table 5.1. Examples of centrally designed formative assessments in LAC Name Country Implementing Implementation Dates Level and scope Institution (during 2021) Dialect (Reading) and Chile Ministry of Education First implementation in 2021 64,333 children tested Diamat (math), in DIA+ in reading and math. EGRA and EGMA, Colombia Carvajal Foundation Since 2017 two times per year 16,000 children tested Tutoring Diagnostic and formative Brazil Ministry of Education (MEC) and the Launched in 2021 Official data not been assessments Center for Public Policy and Educative published yet Evaluation (Centro de Políticas Públicas e Avaliação da Educação, CAEd) Diagnostic Assessments Mexico Public Education Office (Secretaría Launched in the academic 516,000 students from MEJOREDU de Educación Pública, SEP) and the cycle 2021-2022 more than 2,470 schools National Office for the Continuity of (basic education) Learning Improvement (Comisión Nacional para la Mejora Continua de la Educación, MEJOREDU) Independent Learning Mexico Independent Learning Assessments Since 2014 300,000 children Assessments (Medición tested Independiente de Aprendizajes, MIA) Evaluating Breaks Buenos Education office in the Ministry of Since 2016 and annually, 31,400 children in 3rd (Pausas Evaluativas) Aires, Education, Buenos Aires Government except in 2020 and 6th grades tested Argentina in math CREER Project (Proyecto Peru GRADE 2019-2021 (first phase), since Training in 40 schools CREER) 2022 (second phase and (1-4 teachers per implementation) school) Diagnostic Assessment Dominican INICIA Foundation, and Ministry of 2021, applied once 30,000 students tested IQ Platform (Prueba Republic Education diagnóstica de Plataforma IQ) Learning Evaluation Uruguay National Administration of Public Since 2011 (3rd to 6th Open to private and System (Sistema Education (Administración Nacional de grades), annually. In 2014 public schools. de Evaluación de Educación Pública, ANEP) secondary levels are included Aprendizajes, SEA) Source: Tejada et al. (2022). during the pandemic, through the DIA in Chile (see Box availability of this information, paired with practical guid- 5.6), and Evaluar Para Avanzar (Assess to Make Progress) ance on how to use it, can help support students on their in Colombia. Both initiatives provide teachers with instru- learning recovery and safe return to schooling. Mexico ments to be applied to students at the beginning, middle provides another good example through their Indepen- and end of the school year both in academic subjects and dent Learning Assessment (MIA) in Veracruz (see Box 5.4). socio-emotional learning, and the assessment results are combined with protocols, practical guidance, and peda- An especially appealing feature of formative as- gogical support to address the needs identified. Regular sessments is their integration with pedagogy. Two Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 115 Box 5.3. Examples of formative assessments designed at the central level. Brazil, State of Ceará: Diagnostic Evaluations The purpose of this assessment is to diagnose the knowledge and skills of students at the beginning of the school year, so that teachers can use results to plan their classes. Implementation: The assessment is run by the secretary of education of the State of Ceará, Brazil, in coordination with municipalities to administer the assessment and support pedagogical uses at the local level. Mathematics and language tests are aligned with the curriculum, for grades 1-9. The assessment includes printed tests that are admin- istered to the whole class, and individual sessions to measure reading fluency. Teachers from all public schools are required to administer the assessment to their students. Municipalities are responsible for printing and distributing tests to the schools, and for training teachers on how to administer and score them. Use of results: Student and classroom level results are automatically available on an online platform and access to results is granted to educators at the state, regional, department, municipal, and school levels. Workshops offer edu- cators the opportunity to understand, appreciate and reflect on how to use the results. Teachers use results to provide individual student support. At-risk students” receive extra support beyond regular class time. Teachers are provided pedagogical resources and scripted lessons. Results are also used for teacher training purposes. Ceará is considered a model for improving education quality in LAC. Sources: Loureiro, Cruz et al. (2020); Loureiro, di Gropello et al. (2020). approaches can be identified205. One the one hand, initia- arrangements and costs. Ministries/secretaries of edu- tives promoted by ministries or secretaries of education cation, assessment agencies, universities, public-private usually accompany assessment results with pedagogi- partnerships, private foundations, NGOs, and/or civil so- cal guidelines and resources (e.g., teaching and learning ciety have all been involved in the design and implemen- guides, item banks). It is then up to the teachers to make tation of these assessments in LAC. It will be important use of results with the support of these resources. Exam- to keep up with these partnerships and build the right in- ples of this approach can be found in Chile, Uruguay, and centives and capacities to evaluate and scale-up these ini- the state of Guanajuato in Mexico. On the other hand, tiatives. Regarding the costs, it is worth noting that these smaller scale programs and initiatives, led by civil soci- assessments are much cheaper than national assessments ety or others, promote the use of assessment results as for monitoring purposes. Important savings are possible an input to decide next steps in instruction, based on an given that there is no need to hire external administrators, already existing pedagogical intervention. For instance, nor to print and distribute all the tests to schools (schools Fundación Carvajal in Colombia administered EGRA (Ear- or local governments can print them if needed). However, ly Grade Reading Assessment)/EGMA (Early Grade Math- new additional costs arise when integrating the assess- ematics Assessment) to assess foundational reading and ment with pedagogical interventions (e.g., costs related to mathematics skills in some public schools. Results were training teachers to provide focused instruction). used to diagnose student learning and provide focused in- struction using the teaching at the right level approach206. LAC countries should continue the trend towards cen- A similar approach was used by MIA in different school trally designed formative assessments. This is especially municipalities of the State of Veracruz in Mexico (Box 5.4). critical as results from large-scale assessments will not be available for several additional months in many countries, Other important considerations related to centrally de- but also, in any case, as a strong complement to these as- signed formative assessments are their institutional sessments, in the context of a strengthened evaluation 205 Ibid. 206 Ibid. Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 116 Two Years After | Saving a Generation culture where formative and summative assessments Figure 5.7. Different types of learning recovery work together to ultimately improve learning. Formative programs in LAC assessments should keep focusing on the measurement TaRL of basic literacy (e.g., decoding, reading words, sentenc- Teaching at the Right Level Mexico JEE es) and numeracy (e.g., number sense, solving simple re- Extended Teaching Time al-life problems involving operations) skills, while seizing Dominican Republic opportunities to also measure some critical transferable skills. They should also keep stressing teacher participa- tion as a strategy for professional development and em- Early Learners Leveling. powerment. Countries interested in introducing formative Caribbean assessments should explore the best design options for them (e.g., grades tested; tests formats, frequency), tak- Aula Global Tutoring. ing into account their education goals, school culture, and Colombia resources, among others. Above all, they should explore how to integrate these formative assessments with peda- gogical interventions, including learning remediation and recovery programs launched before or as a reaction to the pandemic. Acelera Brasil Acceleration. PAM Brazil Computer-Assisted 5.1.3 Implementing learning Learning recovery programs Uruguay Source: Mancebo and Vaillant (2022). The challenges of the learning recovery are massive. School closures and the reopening of schools chal- lenged countries to innovate in their pedagogical prac- tices. At first, the challenge was more focused on how to should prioritize interventions that already exist and ad- teach effectively in a remote learning environment. With just/improve and scale them up to adapt them to the cir- the gradual reopening of schools, the challenge shifted cumstances. This may also entail generating better data more towards how to recover and support learning in an and improve monitoring and evaluation capacity. in-person or hybrid learning environment. The challenges were and are huge in view of the massive learning losses, Pre-existing learning recovery programs208 received re- which while larger for the most vulnerable students, are newed attention during the pandemic. These programs expected to be very widespread; the need to address the have a long tradition in LAC and refer to education inter- significant heterogeneity in learning levels within class- ventions focusing on students who are lagging behind in rooms; and the need, to a large extent, to recover from their academic learning. They are usually conceived as ad- those losses in the same number of years of schooling. ditional, short-term support that augment regular classes. In general, they focus on foundational skills in language Several approaches and options exist to address learn- (especially reading) and mathematics, with the aim of ing losses. This could include focusing only on the foun- avoiding students repeat classes and/or dropping out from dational skills or lower the level of the whole curriculum; school. Many times, they are fully integrated with assess- have, in some cases, students repeat grades; and/or scal- ments that provide a diagnostic of the initial learning level ing-up learning recovery programs. This section reviews of students, as well as midterm and end-line measures to learning remediation and recovery programs that existed monitor the impact of the program. These programs adapt before the pandemic, as well as those programs intro- and personalize the curriculum to the learning needs of duced as a reaction to it, building on a paper specifically the students, emphasizing teaching at the right level. They undertaken for this report,207 to derive useful lessons for are mainly offered in person, but the pandemic pushed for introduction and scaling-up. Wherever possible, countries the adoption of remote strategies too. Figure 5.7 presents 207 Mancebo and Vaillant (2022). 208 Strictly speaking programs introduced before the pandemic were more about remediating/catching-up than recovery, but we use the term “learning recovery” programs to be more comprehensive and aligned with the most recent programs and challenges. Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 117 Box 5.4. Example of a learning recovery program: Mexico. Teach at the Right Level (TARL)/ Medición Independiente de Aprendizajes (Independent Learning Assessment) (MIA) Since 2016, MIA has served near 13,000 students from 27 municipalities of the state of Veracruz in Mexico and works with over 80 civil and educational organizations. To support sustainability of the intervention, MIA puts strong em- phasis on working with local governments, building local capacity, and offering technical support. Evidence shows significant increases in learning scores when comparing pre- and post-intervention results. MIA supports the development of foundational skills in mathematics and language, as well as some transferable skills (i.e., self-care, civics, socio-emotional). It uses two main strategies: • Household measurement of students’ foundational and transferable skills at the beginning and at the end of in- terventions, led by independent volunteers from communities • Short and scalable pedagogical interventions based on the principles of “teaching at the right level”. MIA offers different types of interventions, including regular school or extra-curricular activities, summer camps and clubs for kids, as well as workshops for parents to provide learning support for their kids at home. Beginning of intervention Middle of intervention End of intervention • Diagnostic assessment for kids from • Reading intervention: 20 sessions of • Final assessment for kids at grades 3-6. grades 3-6. 90 min. each. • Sharing results and end of the • Sharing results and workshop with • Math intervention: 20 sessions of 90 intervention. parents. min. each. • Citizen-led actions to support • Grouping of students (20 max.) by • At least 2 sessions per week. sustainability of the intervention. learning needs Source: Own elaboration on the basis of information from MIA’s website (https://www.medicionmia.org.mx/). pre-existing flagship programs in LAC and Box 5.4 provides • Tutoring:  Individual or small group interventions pro- further details on one of these programs. Annex 5.1 pro- vided by one person to one student or to a small group vides a global snapshot of such programs beyond LAC. of students. E.g., Aula Global, Colombia Despite their commonalities, learning recovery pro- • Teaching at the right level: I nterventions that group grams can be very heterogeneous in their aims, design, students based on their actual skills or learning lev- and implementation features. Six main (non-exclusive) el in order to provide targeted instruction. E.g., TaRL/ categories can be distinguished209: MIA, Mexico. • Catch-up:  Interventions targeting students who need • Extra instructional time: Extension of the instruction- to recover specific content, skills, competencies in or- al time by providing extra-curricular activities. E.g., der to catch up and benefit from regular instruction Jornada Extendida, Dominican Republic. in the classroom. E.g., Early Learners, the Caribbean. • Computer based programs: Interventions that pro- • Accelerating programs: Interventions targeting over- vide more personalized instruction with tailored or aged students or students who dropped out and have adaptative educational programs, learning games, returned or are planning to go back to school. E.g., activities or the like. E.g., PAM, Aprendizaje asistido, Acelera Brasil. Uruguay and Khan Academy, Brasil. 209 Mancebo and Vaillant (2022). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 118 Two Years After | Saving a Generation Figure 5.8. Number of LAC countries that implemented wide remedial measures to address learning gaps when schools reopened after the first closure in 2020, by type of remedial measure 10 9 8 8 7 7 6 6 6 6 5 5 5 4 4 4 4 3 3 2 0 Focus on reducing Focus on students Focus on students Focus on students with Focus on students learning gaps unable to access distance at risk of drop-out or certi cation exam transitioning cycles learning grade repetition Primary Lower Secondary Upper Secondary Source: UNESCO, UNICEF, WBG, and OECD (2021b). Survey on National Education Responses to COVID-19 School Closures, round 3 (Feb-Apr 2021). Overall, 26 LAC countries answered the survey. Another key feature of learning recovery programs is learning, especially among lower-income students.213 An- who provides the pedagogical intervention. In some nex 5.1 summarizes information about flagship programs cases, the person in charge could be an active teacher in and outside LAC that existed before the pandemic, and from the school, a future teacher or a retired one. In oth- related evidence to substantiate their impact. ers, students from the same school provide assistance to other students. It is common to see remedial programs LAC countries also responded to the pandemic by benefiting from parents or other community members. launching new “emergency measures” for learning re- Interventions may rely on paid staff or volunteers to op- covery. Many of these initiatives built on the same prin- erate. Tutors are typically used to provide one-on-one in- ciples and approaches of already existing programs. As struction or instruction to small groups of children with shown in Figure 5.8, the most common measures wide- similar learning needs. A concern, though, is the little or ly implemented when schools reopened were: remedial no training these people receive, both to provide academ- measures with a special focus on students in upper sec- ic and emotional support. This is a critical constraint giv- ondary with a certification examination; remedial mea- en the deficit of qualified teachers in LAC and working sures to reduce student learning gaps in primary and conditions that make it hard for them to get additional lower secondary education; and remedial measures with training.210 a special focus on students who were unable to access distance learning in primary education. Several of these There is evidence of the positive impact of some of the programs were introduced after regular school time or flagship learning recovery programs in LAC and beyond. during scheduled school holidays. Brazil, Cuba, and sev- In Colombia, students participating in “Aula Global” showed eral English-speaking Caribbean Islands are among the significantly better results when comparing achievement countries that reported implementing wide remedial scores before and after the tutoring intervention, and measures in different school cycles. when comparing results from the intervention versus con- trol group. Students in the intervention group were also Despite a learning crisis on the continent, programs less likely to repeat grades and to drop out from school211. launched with the pandemic seem to remain on the “Teach at the right level” programs have proved cost-effec- margins of education systems. They may exist as rel- tive to boost learning in developing countries such as Chile, atively small and short-lived initiatives or projects and Mexico, India, and Kenya.212 The use of the computer-based many do not go beyond a pilot or initial phase. They programs in Uruguay has also proven effective to boost may be located in a particular school or network of 210 Ibid. 211 Mancebo and Vaillant (2022). 212 J-PAL (2019). 213 Perera and Aboal (2019). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 119 schools or serve the students within the boundaries of Figure 5.9. Key questions for countries to consider a municipality or school district. Or their reach may be when implementing and scaling-up learning recovery constrained by limited access to technology and con- programs nectivity issues. Does it make sense? Institutional arrangements may be weak and local in • Is there a clear theory of action of how the program will support learning? nature, further threatening the scalability and sustain- ability of these emergency programs. For instance, when • Is the program a good fit for my country given school culture, values, practices? projects only exist within a school or within the boundar- • Is the program disruptive but within the boundaries of ies of a municipal or district office of education, or when acceptable? they are led by civil society. This institutional fragility may be also observed in the day-to-day operations of some Does it work? projects. It is common to see little coordination between • Is the program effective on reaching its objectives? the remedial program and the regular school activities. • What is the evidence of program impact? Teachers may not even know there is a learning recovery program in their schools and, if they do know, they may • Do students, teachers and parents like the program? perceive it as a threat. Why is someone else teaching my kids? Why are they using a different pedagogical approach How much does it cost? than the one I use in classes? • How much money needs to be disbursed? • Is the program relatively cheap compared to regular A key question that remains is how effective these new operations or new interventions? programs have been to respond to the pandemic. While • Is the program cost-effective? these interventions are highly valued from a conceptual point of view (i.e., theory of action, personalized pedagog- Is it manageable? ical approach), the evidence to substantiate their impact • Are the institutional requirements to implement the program is limited. When COVID-19 hit, there was little time for de- reasonable? signing the interventions themselves, not to mention for • What team is needed to implement the program ? designing any program evaluation. Without any baseline, • What and how much training do teachers and tutors need? midline or end line data, it is hard to gauge their impact • What physical infrastructure is needed? on learning. • What technological infrastructure is needed? Lessons learnt Both well established and new programs launched Countries willing to implement learning recovery pro- during the pandemic offer a promising path for recover- grams at scale will need to pay attention to critical ing from learning losses in the short-run and improving questions and features which are more or less condu- learning in the longer-run in LAC. Their more personal cive to scalability. Figure 5.9 elaborates on those ques- approach to learning, their focus on the overall well-being tions and Table 5.2 highlights such critical features for the of students, and their greater flexibility to operate beyond flagship programs reviewed. school boundaries, are all desirable features that should stay beyond the pandemic. To sum up, countries must urgently step-up initiatives and programs to recover from learning losses and ac- The key question is how to do it. How to provide person- celerate the learning process, building, as much as alized pedagogy in the context of mass education? How to possible, on existing programs, evidence and les- deal with constraints such as shortages of qualified teach- sons. The focus of these programs should primarily be ers, institutional capacity, and limited human, physical on foundational literacy and numeracy skills, including and financial resources? How to transfer and adapt pro- decoding and reading simple words; and number sense grams from one country to the other? How to ensure their and operations. To maximize learning, especially in the scalability and sustainability? earlier grades, instruction should be provided in the Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 120 Two Years After | Saving a Generation Table 5.2. Critical features needed to adapt, scale-up, and ensure sustainability of flagship learning recovery programs Type of program Requirements Pedagogical Model Human Resources Institutional and other aspects Catch-up Curriculum Identification of necessary • Accurate definition of the program’s institutional aspects adaptation. human resources • Definition of the role of each of the participants • Clear definition of the objectives of the program Training for teachers in charge of the catch-up groups. • Diagnostic assessment tests • To group students according to their learning level • Evaluation of the program • Financing Accelerate Understanding the Identification of necessary • Accurate definition of the program’s institutional aspects model “Learning human resources • Definition of the role of each of the participants acceleration”. • Clear definition of the objectives of the program Teachers’ training with focus on the “Learning Acceleration” • Diagnostic assessment tests model. • To group students according to their learning level • Evaluation of the program • Financing Tutoring Designing strategies Identification of necessary • Accurate definition of the program’s institutional aspects to support a human resources • Definition of the role of each of the participants personalized • Clear definition of the objectives of the program pedagogical Teachers’ training with focus on personalized pedagogical • Diagnostic assessment tests support. support strategies. • Evaluation of the program • Financing Teach at the Designing strategies Identification of necessary • Accurate definition of the program’s institutional aspects Right Level of a personalized human resources • Definition of the role of each of the participants (TaRL) pedagogical • Clear definition of the objectives of the program support. Teachers’ training with focus on personalized pedagogical • Diagnostic assessment tests support strategies. • To group students according to their learning level • Evaluation of the program • Financing Extended Designing Identification of necessary • Accurate definition of the program’s institutional aspects Pedagogical strategies for extra human resources • Definition of the role of each of the participants Time pedagogical time • Clear definition of the objectives of the program usage. Teacher training for classroom and other roles (physical • Diagnostic assessment tests education, art, ludic activities, • Infrastructure availability for example). • Evaluation of the program • Financing Training on usage of extra pedagogical time. Computer- Availability of self- Identification of necessary • Clear definition of the objectives of the program assisted learning computer human resources • Availability of IT infrastructure (hardware, software and learning programs with connectivity) program exercises, feedback Classroom teachers and • Evaluation of the program and personalized assistants’ training • Financing support. Assistants’ training focused on: personalized assistance to students according to the chosen platform. Classroom teachers’ training focused on: digital literacy with especial emphasis to the chosen platform. Source: Mancebo and Vaillant (2022). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 121 mother tongue of the students214. To recover learning put special focus on pre-primary education to make sure losses, these programs need to start by diagnosing the recovery starts early (Box 5.5). actual learning level of students, and to keep monitoring it during the intervention. Good monitoring and evalua- tion systems and strengthening institutional capacities to implement at scale will be critical. Teachers, tutors, 5.2 Addressing psychosocial health parents, and other implementers need to be trained on and well-being how to target instruction, by teaching at the right level of students. This would require training on how to inter- Psychosocial health and well-being of students re- pret diagnostic results and how to adjust pedagogy to quire explicit support and prioritization during school the real learning needs of their students. Priority should reopening and the recovery phase. Despite the limited be given to hands-on, continuous training together with data, the global nature of the pandemic has allowed to teaching guides and pedagogical resources for teachers. realize that many students have experienced emotional Targeted instruction has strong potential to also improve distress over the last 2 years. The previous chapters doc- learning in the longer run in LAC.215 While information umented a deterioration of mental health and well-being on unit costs is unfortunately missing for most type of of children and youth across education levels in several programs, based on critical features to adapt and scale- countries. Even before the pandemic, schools played a key up and evidence of results, programs like “Teaching at role as a platform for providing other essential services the right level” may be especially promising to both re- such as school feeding programs, health services and psy- cover and accelerate (as illustrated by the Mexico and chosocial support, in addition to education. As schools re- India cases – also see Boxes 5.1 and 5.4). Other pro- open, focusing on the psychosocial health and well-being grams, however, such as very structured “catch-up” and of students will, on the one hand, have positive effects on “tutoring” programs may be lighter in terms of required attendance and retention, and on the other hand, facilitate training. As pedagogy improves, it will also be critical to learning. maintain an alignment between pedagogy, assessment, and curricula. A comprehensive response package is needed. In align- ment with the recently published RAPID framework218, Beyond these specific interventions, a focus on early core actions to be included in the package could comprise: childhood education will be essential in supporting (i) providing students with nutritious meals; (ii) guaranty- readiness to learn, acquisition of transferable skills ing adequate water, sanitation, and hygiene services; (iii) and socio-emotional growth in both the short and lon- ensuring students’ safety through appropriate risk screen- ger-term. Pre-primary education provides children with ing and protection services; (iv) giving students spaces to cognitive stimulation and socio-emotional development, play; and (v) assure access to critical mental health and which prepares them for their future course of learning psychosocial support. Besides meeting students’ special and their life trajectory. Children who lag behind in learn- needs, it is also important to train teachers and school ing during the early years are found to stay behind for the leadership teams on psychosocial support. These actions remaining time they spend at school.216 Emerging evidence could be complemented by the collection of real-time in- from the current crisis also suggests that pre-primary edu- formation on the current psychosocial needs of students cation may have a protective effect against learning loss.217 to establish specific actions and support during the recov- As countries recover, it is therefore also critical that they ery phase. 214 WBG (2021d). 215 That is a pedagogy that provides more tailored and sustained support for students that are at different learning levels. Even before the pandemic, there was a need to tailor instruction to the diverse knowledge and skills of the students, and to the needs of students who were not learning at school. The rationale for im- plementing remedial programs in LAC should change from exceptional, short-term interventions towards the provision of a more permanent support during regular class hours. The shift caused by the pandemic and the programs implemented to recover from the learning losses are providing an opportunity for “normalizing” some remedial programs that can be scaled up to a personalized approach to education. 216 Nugroho et al. (2021). 217 Ibid. 218 UNICEF, UNESCO, and WBG (2022). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 122 Two Years After | Saving a Generation Box 5.5. Recovering and accelerating in ECE Evidence has shown that children’s access to high-quality ECE benefits early skill development as well as lon- ger-term physical and mental health, educational attainment, and earnings219. Existing evidence from different disciplines has confirmed that what happens during the first years of life has life-long consequences. Positive and negative experiences in early childhood (before the age of 8 years) have long-term implications with respect to phys- ical and mental health, life-long learning and positive participation in their communities and society. The pandemic seems to have had a strong impact on ECE education where children from disadvantaged con- texts appear to be the most affected. According to McCoy et al. (2021), at least 167 million children have lost access to early educational support. At the same time, low-income countries tend to be the most affected. In 2020, high-in- come countries reported an average of 46 pre-primary instruction days lost due to COVID-19, while low-, lower-mid- dle and upper-middle income countries respectively lost 90, 122, and 106 days220.    Beyond the long-term impact on learning, the closure of early education facilities and limited interactions with extended families also deprived children from multidimensional benefits. School closures have led young children to (i) miss out opportunities to experience rich and sustained social interactions that promote growth, de- velopment, emotional regulation, and learning; (ii) lose access to critical services such as the provision of varied, nu- tritionally balanced, and quality food, the delivery of routine immunization, and the early detection and notification of risks for children; and/or (iii) miss protective environments provided by ECE centers for children (and other family members) living in a family context with physical or emotional violence. Young children have been exposed to mul- tiple risks over the past few years, including few and poor learning opportunities, which may pose significant risks to their present and future development. There are, however, evidence-based interventions that can be promoted to reverse this challenge.  There has been less government response to the pandemic’s effects on children in pre-primary education (who, at the same time, were less likely to have options for learning remotely while their schools were closed). Few government guidelines covering remote-learning responses to COVID-19 closures refer to pre-primary education spe- cifically221. Only 60 percent of countries had digital and broadcast learning available for pre-primary schools, while almost all countries (more than 95 percent) had these options available for primary and secondary levels. Fewer than one in five Lower Middle-Income Countries (LMICs) reported that over 75 percent of pre-primary students took part in remote learning (Figure B.5.5.1).222 Therefore, the call for action is in specific areas to address ECE challenges in the region and prepare families and early childhood education systems to deal with the low levels of school readiness of new children entering the school system post-COVID-19. UNICEF has called for 10% of both domestic financing and international aid in education to be allocated to pre-primary education223. Protecting and continuing to increase pre-primary education budgets to reach this target is important to allow countries to respond and recover from COVID-19 learning losses. Every dollar (US$) spent on pre-primary education is estimated to result in US$ 9 of benefits to society224. Moreover, there are short-term recommendations in other specific areas: (i) Reopening ECE Services; (ii) Recovering Normality: Mental Health and Well-Being of Children and Caregivers; and (iii) Recovering Lost Learning.  (Box continues on next page) 219 McCoy et al. (2021). 220 Ibid. 221 Nugroho et al. (2021). 222 Ibid. 223 Ibid. 224 Ibid. Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 123 Box 5.5. Recovering and accelerating in ECE (continued) Figure B.5.5.1. Pre-primary students in LAC: Remote learning during school closures, and support received following reopening, 2021 Assessment of gaps in students learning that may have accumulated during school closures widely implemented a er reopening (n=43) 33 Remedial measures to reduce student learning gaps widely implemented following reopening (n=43) 44 More than 75 percent of students attended school in-person a er the reopening of schools (n=85) 26 More than 75 percent of students followed distance education during school closures (n=85) 18 0 10 20 30 40 50 60 70 80 Upper Secondary Lower Secondary Primary Pre-primary Source: Nugroho et al. (2021)  1. The reopening of presential service provision and ensuring that children receive supportive and nurtur- ing learning opportunities is an urgent priority.  Closed early childhood services and schools should be re- opened immediately, particularly since COVID poses limited health risks for this population and since many of these early skills are not adequately developed in the home. Not only will this help return children to learning, as well as stimulating environments that promote their integral development, but it would also help women return to the labor market. We make an urgent call to governments and public and private service providers in the Latin America and Caribbean region to take all necessary measures for the safe reopening of early child- hood care centers and preschools. 2. If measures restricting in-person attendance persist, it is critical to ensure the need for improved access to Internet and mobile phone services to enable the provision of remote or hybrid learning opportunities. Additionally, it is important to strengthen the implementation of programs and strategies that ensure learning continuity for children, especially for those in situations of extreme vulnerability, with special educational needs, or living in isolated or rural areas (e.g., virtual instruction, television, and radio shows, family-friendly apps). Ef- forts by governments, agencies, civil society, and relevant partners to support young children and their families will be critical, for example guaranteeing that remote-learning strategies are complemented by services that support parents and caregivers to provide warm, stimulating care, while also ensuring their own mental health and financial security225. At the same time, as pre-schools reopen, pre-pandemic subsidies for childcare as well as nutritional supplementation should be restored as well.   (Box continues on next page) 225 McCoy et al. (2021). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 124 Two Years After | Saving a Generation Box 5.5. Recovering and accelerating in ECE (continued) 3. Focus on behavioral skills that help children later in life.  A setting of stability and tranquility in pre- sential early learning programs to reduce stress for children is the most critical objective, even more than recovering academic/cognitive skills. Children may have experienced traumatic experiences such as bereavement or illness of a loved one, housing or food instability, parental unemployment or financial bur- den, distancing from family and community members, or ongoing toxic stress such as violence in the home / domestic violence or various forms of abuse exacerbated by confinement. It is important to destigma- tize conversations about all aspects of well-being, including mental health and self-care. The ECE workforce should be supported in this effort with tools, trainings, and resources. They can also play an important role to invite members of the community to express their needs, successes, and challenges to establish a sense of stability and feeling well. These models support children’s socio-emotional awareness and encourages their own expression.   4. Early childhood services and schools need to provide parents and caregivers with as much support as possible, making information as well as resources and capacity building opportunities available to them. Establishing helplines to foster dialogue can help identify the concerns of parents and caregivers during and after the pandemic and regular communication helps to reduce fear, to encourage children’s return to early childhood centers and to schools, and to make it easier for parents and caregivers to understand and implement protec- tion measures226. During the pandemic, the Bank and others have funded trials of remotely delivered parenting programs, and some have generated impact. For example, an impact evaluation funded by the Bank in Guate- mala found that parenting content delivered through voice messages sent to cell phones improved children’s vocabulary and increased parents’ participation in play activities. At the same time, this light-touch intervention improved caregiver mental health.  5. Governments and partners should introduce accelerated, bridging, or remedial programs. Successful case studies from the following three types of programs include: (i) Accelerated school readiness programs that usu- ally take place in the school holiday period, prior to Grade 1 in primary school; (ii) Bridging programs that take place in the first months of Grade 1; (iii) Remedial programs that take place in the first two years of primary school. These types of programs have been used successfully to provide transitionary pre-primary education for children who would otherwise not be able to access it at all. They can have multiple benefits: ameliorate the long-term impact of school closures on pre-primary-aged children directly affected by COVID-19; contribute to the expansion of the pre-primary education provision; and strengthen the resilience of the education system in facing future crises.  The analyses of successful programs (as discussed above) provide useful lessons learned for each of the 5 ar- eas of the conceptual framework for ECE. The Reopening of ECE centers and pre-schools, along with the priority to ensure efficient recovery and to ensure future resilience provides an opportunity to strengthen and re-imagine ECE systems (quality, equity/inclusiveness, resilience/sustainability) within the following ways:  Planning and budgeting, curriculum implementation, workforce development, parents and community engagement, and quality assurance.    Sources: McCoy et al. (2021); Nugroho et al. (2021). 226 UNICEF LACRO (2020). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 125 There have been some examples in the region of inter- psychologists, orientation teams, teacher mentors or ventions to support psychosocial health and well- be- coaches can be a powerful tool to ensure teachers are sup- ing. Chile provides a good example of a balanced learn- ported and have the adequate tools to support students. ing and socio-emotional recovery package (Box 5.6). And Some of the services provided may also require collabo- similarly, also does the State of Guanajuato in Mexico with ration with other governmental sectors such as health and his comprehensive recovery package in the context of its child protection. Social Pact for Education (as seen in Box 3.3). Of note is also the new Ecuador multi-pillar plan and strategy to pro- mote learning recovery “Learning on Time” (Aprender a Ti- empo)227. The plan has four key pillars: 1) socioemotional 5.3 Addressing digital gaps support, 2) prioritization of reading and math, 3) strate- gies to prevent school dropout, and 4) recovery of aca- The availability and quality of available infrastructure, demic skills. However, one constraint for the design and lack of student and teacher digital competencies, and implementation of psychosocial recovery packages is still institutional constraints, have been hindering the effec- the limited hard evidence available on the effectiveness of tiveness of remote and hybrid learning and constrain- various interventions, which also flags the importance of ing digital innovations. As shown in Chapter 1, a central improving good data for decision-making. aspect limiting the success of virtual and hybrid learning in the region, and which also affects other efforts at sup- As they support students, it is essential that the psycho- porting digital innovations in education, is that access to social needs of teachers be also part of the conversa- and quality of the Internet services is still very unequal. tion. It is time to start gathering data on teachers’ partic- Internet access and quality is also an issue in schools, ular challenges and expectations during school reopening. where it can constrain recovery and acceleration. In Chile, The pandemic has introduced new sources of job-related almost 1 out of 3 schools does not have Internet access.229 stress for teachers as education systems have been navi- Lack of proper equipment and digital skills (e.g., lack of gating between remote and hybrid education. In addition knowledge on how to use digital platforms) are essential to this, teachers, 70% of which are female, have had to constraints affecting the effectiveness of hybrid learning, disproportionately face additional stressors such as taking and more broadly the use of technology in education.230 care of their own children at home, and household admin- As seen previously, lack of adequate digital skills was an istration. As these teachers under stress provide support important pre-pandemic challenge for both students and to students, their personal well-being becomes paramount. teachers, and so have been institutional constraints. Portals and platforms with support resources and instru- ments and surveys that can gather information on teach- LAC countries have begun implementing policies and ers’ preferences and expectations are needed across the programs to reduce the digital divides, but efforts re- region. Peru provides an example of both (Box 5.7). quire higher and sustained investment. Some LAC countries invested in improving Internet connectivity in Finally, it will also be important to identify and align schools and/or providing subsidized Internet access to other actors in the education system, and beyond, to families. In Uruguay, “Ceibal en Casa” made an agreement support students and teachers. Beyond teachers, school with the National Telecommunication Agency (ANTEL ac- counsellors, school management teams, local and region- ronym in Spanish), the state national Internet provider al authorities can also play important roles support the and market leader, and Claro, a private Internet provid- well-being of students and teachers upon returning to er, trying to connect and provide resources to all students, school. School counsellors for instance could play a crit- especially those most vulnerable. For instance, the Gov- ical role, even more relevant for low-achieving or low-in- ernments of Argentina, Peru, Chile, and Ecuador made sig- come students.228 Now more than ever, aligning school nificant investments to improve equipment and Internet 227 See https://educacion.gob.ec/aprender-a-tiempo/. 228 Mulhern (2020). 229 Del Castillo (2021). 230 In order to enable countries to identify good practices and areas where EdTech policies can be strengthened and monitor progress as countries take action, the World Bank, with the support from Imaginable Futures, has created the EdTech Readiness Index (ETRI). This tool will increase learning opportunities and reduce in- equalities while investments in EdTech are done. The index is organized around 6 pillars: school management, teachers, students, devices, connectivity, and digital resources. The first three pillars are the actors in the education system, while the last three are the inputs and infrastructure that the actors need to use EdTech. Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 126 Two Years After | Saving a Generation Box 5.6. The strengths of the recovery response in Chile To support schools during the reopening, the Ministry of Education of Chile developed the “Chile recovers and learns” (Chile se Recupera y Aprende) strategy, implemented since April 2021. The strategy was articulated around: (i) keeping and attracting students back to school; (ii) recovering and leveling learning; and (iii) socio-emotional well-being, as described below. While hard evidence of impact is not yet available, survey results (from the 2022 Encuesta de Mon- itoreo Nacional del Sistema Educacional) have showed overall high levels of satisfaction with these measures (Figure B.5. 6.1). Figure B.5.6.1. Evaluating key recovery policies for 2022 Curricular Priorization 1 2 7 39 51 JEC Flexibilization 3 3 6 33 56 Socio-emotional well-being 13 11 43 42 kits for students Learning recovery pograms 1 3 9 44 43 High speed connectivity 1 2 3 24 70 DIA Assessments 2 4 9 38 47 Early Warning System (SIGE) 2 6 15 42 35 Escuelas Arriba 6 8 18 37 32 0 10 20 30 40 50 60 70 80 90 100 % of Establishments Not useful Slightly useful Moderately useful Very useful Extremely Useful Keeping and attracting students back to school a. SMS/WhatsApp campaign. Parents of tutors of out-of-school children were contacted and encouraged to send their children back to school. b. Early warning system. The Ministry of Education implemented a nationwide early warning system to alert schools of the 10% of students with the highest risk of dropout. Schools could then mitigate dropout with target- ed and timely support to students at risk. Recovering and leveling learning a. Diagnostic. The Education Quality Agency (Agencia de Calidad de la Educación) developed a set of “comprehen- sive learning diagnostic” (Diagnóstico Integral de Aprendizajes – DIA) tools to assess aspects related to well-being of students and socio-emotional skills, as well as students’ learning in math and language. Tools were delivered to schools to be used three times over the school year (March, August, and December). These instruments helped produce reports with results at the classroom and school levels. Schools received orientation on how to use the tools and the diagnostic results. The DIA diagnostic increased from a coverage of 20% of students in 2020 to 80% in 2021. (Box continues on next page) Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 127 Box 5.6. The strengths of the recovery response in Chile (continued) b. Curricular prioritization. The curricular prioritization identified the main learning goals for each grade and gave flexibility to schools for implementing a study plan aligned with their methodologies and internal organization. c. Learning recovery. Schools were able to enroll in a plan for recovering learning, where they received support in pedagogical practices and resources to recover learning, a set of guidelines for teachers and students, and continuous monitoring and feedback. Teachers received training and the Ministry of Education monitored closely with in-field visits. Socio-emotional well-being Each school received a kit of resources to support teachers and students to promote positive relationships and school climate during school reopening. Source: MINEDUC-Chile and WBG (2022). Box 5.7. Tools and surveys for teachers’ well-being in Peru • Peru implemented the strategy “Te Escucho Docente (I Listen to You, Teacher)”, a portal where teaching staff can access resources and support to address their socioemotional needs that includes materials to cope with stress or anxiety management in response to the health emergency and remote work. • Every two years the Peru’s National Teacher Survey (ENDO) collects information on teachers on their sociode- mographic and socioeconomic characteristics, their professional career, their perceptions regarding the working conditions that increase their well-being and, on the policies and programs that impact their work, as well as on and future expectations. their values ​​ Source: Own elaboration on the basis of information from the Ministry of Education of Peru. connectivity for primary, secondary, and technical schools further ramped-up and sustained. IAD (2021b) shows that nationally (Table 5.3). Brazil implemented a new pro- $47 billion is required to connect the region to the Inter- gram (Programa Inovação Educação Conectada, PIEC) to net for educational purposes by 2030. Overall, it will also support the universalization of public schools’ access be critical to ensure that investments in infrastructure are to broadband Internet and to promote the use of digital not only for emergency but for the “long term” to support technology in basic education.231 Moreover, LAC countries learning acceleration and resilience. provided students and teachers from vulnerable popula- tions with devices such as computers, smart phones, or As seen previously, a positive effect of the crisis was tablets to address the digital divide. Countries like Argen- the pro-activeness supporting teachers and students’ tina, Ecuador, Panama, St. Lucia, Saint Vincent, and the digital skills development during the pandemic. As Grenadines engaged in large procurement contracts to schools return to normal, it would be relevant to under- buy and distribute thousands of computers and tablets stand the way digital skills were applied during remote to teachers and students. Efforts however will need to be teaching, so as to plan relevant strategies for future 231 The program started in 2017 with the goal of providing connectivity to at least 44.6 percent of students, and in 2019, this target was revised to 85 percent of students. The PIEC, which entered its third stage in 2022, now aims to reach 100 percent students in basic education by 2024. Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 128 Two Years After | Saving a Generation Table 5.3. Selected initiatives to improve Internet connectivity for schools and students in LAC Country Program Budget Program scope Ecuador Campaña “Conectando al Futuro” - To reduce connectivity gaps of teachers and students by providing them with Internet access and tablets, prioritizing those living in rural areas. Perú Estrategia “Todos y todas conectados” USD 50.7 million* The objective is to give connectivity to 18,000 schools guarantying access to the TV learning program “Aprendo en casa” implemented by the government. Chile Conectividad para la “Educación 2030” USD 16 million** To give high speed Internet access to more than 10.000 educational institutions (around 3.273.00 students) by 2029. Argentina “Juana Manso” USD 21 million To give students access to digital technology and devices (i.e. 70.000 computers). Source: UNESCO (https://siteal.iiep.unesco.org/respuestas_educativas_COVID_19). Different sources. *This amount corresponds to USD $14 million for the Inter- net access strategy and $USD 36.6 million included for the National Telecomunication Program (Pronatel) in 2021. **This amount corresponds to the upper limit per year. training. For instance, one recent survey among teachers interaction with digital devices and solutions. Digital in Greece showed intense use of digital tools for finding competencies frameworks for teachers and for students and developing educational resources, but little to no use should be supported and so periodic assessments of of such tools for student assessment or for feedback pur- digital skills. Digital literacy needs to include capabilities poses -key features of a meaningful teaching-learning cy- such as critically consuming and producing digital con- cle232. Several European education systems, like Portugal, tents, or problem-solving skills to both systematize and Belgium, Estonia, and Czechia, have invested recently in decompose a problem and formulate strategies to choose specific digital training strategies for teachers233, and most an adequate solution236. Definitions of digital literacy have OECD education systems purposely included teachers’ covered different domains, from general capabilities for digital skills development as a core area in their recovery learning, living, and working in a digital society237 to the plans (Belgium, France, Germany, Greece, and Spain234). more complex definitions used by UNESCO, the Europe- Furthermore, robust measures of digital skills are not only an Union or London School of Economics/International needed for better design and implementation of blended Telecommunication Union, that include evaluation and learning policies, which will remain very relevant in the creation of digital content, communication, online safety post-pandemic, but also for better planning of the future skills, self-control in the use of online resources, the abil- trajectory of students. Recent evidence points out at the ity to distinguish credible and high quality contents, the increase in the demand of digital skills for different sec- understanding of which technology is more appropriate tors. Even though this trend preceded the pandemic, a for each objective, among others. The complexity implied business survey in 2021 that involved various regions and by digital literacy, and the range of skills and tasks that companies showed that the prioritization of digital abili- it involves, highlights the relevance of adopting digital ties was 16% higher than in 2019235. competencies frameworks for teachers and for students, that define minimal levels of competence, at least at the A more comprehensive understanding of digital lit- national level, to regularly monitor level of performance eracy must go beyond the abilities needed for the and collect evidence for design and re-adequate contents 232 Perifanou et al. (2021). 233 European Schoolnet (2021). 234 Zancajo et al. (2022). 235 Feijao et al. (2021). 236 Alexander et al. (2016); Van Laar et al. (2020). 237 JISC (2014). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 129 and strategies. Several national and regional examples Technology could also help develop hybrid learning mod- exist already (even in the region, where Chile and Colom- els to increase access and expand the school day in effi- bia have developed national competencies framework for cient ways. The recent Digital Equity Law just approved teachers) that could be adapted to each national context in Panama (Box 5.8) is an excellent example of a strong and policy objectives (See Box A.5.2 in Annex 5.2). Equally legal and institutional framework to enable these and oth- important will be to ensure that digital skills are also reg- er large-scale innovations. During the closure of schools ularly assessed though standardized/common regional in- and the recovery phase, Panama has been a leader in the struments in the field of digital skills. A repeat of the ICILS region in the design of a long-term vision on the role of should be encouraged. education technology to accelerate learning and reduce learning gaps between poor and vulnerable students and Assuming basic connectivity and digital skills gaps are those from better-off socioeconomic backgrounds. addressed and that the right incentives and capacities are in place to make use of the data and technology in- novations, there is potential to build on data and tech- nology to improve service delivery through strength- 5.4 Key takeaway messages ened information systems, platforms to enhance targeted learning or, yet, innovative and sustainable Addressing the massive learning losses requires action uses of remote and hybrid learning. As well illustrated in on three critical interrelated pillars. School systems usu- recent reports,238 leveraging efficiency-enhancing techno- ally foster learning based on three main pillars or dimen- logical and data innovations, introduced before or during sions: a curriculum that states what students should learn; the crisis by some countries, could be especially instru- a pedagogy that guides how to teach towards curricular mental in improving service delivery. Improving access to, goals; and assessments that provide evidence on student quality and timeliness of data, including through stronger learning. These three pillars were greatly disrupted during Education and Management Information Systems (EMIS) the school closures, in a context of very challenging remote and monitoring and evaluation systems, can have multiple learning experiences, leading to large learning losses. It applications, from supporting school reopening process- is imperative for all countries of the region to frame their es, to student and outcome monitoring and efficient allo- learning recovery process around these three pillars, build- cation of human, physical and financial resources across ing, when possible, on measures and programs pre-existing areas and schools. Early warning systems reviewed pre- the pandemic or newly introduced, while addressing pre-ex- viously are a good example of information used to mon- isting institutional constraints and service delivery failures. itor student attendance and frame support interventions which could help improve the efficiency in the use of re- Curriculum should be consolidated with a focus on sources. As seen above, comprehensive assessment sys- foundational skills. The pandemic opened the door for tems to monitor and support student learning and well- countries to make substantial revisions to their curricu- being, and to timely monitor and evaluate programs and lar policies, prioritizing what was considered essential interventions, are critical to improve student learning and learning for all students. Teaching mathematics and lan- scale-up what works. Comprehensive information systems guage (reading, writing, literature) became a priority over can also include invaluable information on student gradu- other subject areas in several countries. Countries should ation and employment rates, by integrating tracer and la- keep prioritizing foundational skills in the short and lon- bor market surveys, essential tools to track the quality and ger-term, while integrating transferable skills as a neces- relevance of secondary and tertiary programs. Technology sary complement. is also at the core of computer-based adaptive learning programs, illustrated above through the cases of the PAM Learning assessments should be urgently re-priori- platform in Uruguay, as well as in Ecuador, where an adap- tized. Data was and is needed to diagnose the magni- tive computer-assisted remediation program, supported tude of learning losses and to take adequate actions. Most by the World Bank, has been used to support students in large-scale national assessments were cancelled in 2020. technical and technological institutes239. While very prom- Most resumed in 2021, but with results not yet available. ising, bringing to scale these programs is still a challenge. Teacher led formative classroom assessments had to be 238 Including WBG (2021a). 239 Angel-Urdinola (2020). Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps 130 Two Years After | Saving a Generation Box 5.8. The Long-term vision of the role of education technologies in accelerating learning and reducing learning gaps in Panama On April 6, 2022, Panama approved a far-reaching law called the Digital Equity Law (Ley de Equidad Digital). This law establishes the general guidelines for the formulation, development, and implementation of education public poli- cies aimed at increasing the equity of the education system. To achieve this, the law proposes the implementation of pertinent teaching-learning models, including the delivery of flexible modalities for the education services, facilitat- ed by the technological transformation, which guarantee the delivery of creative and innovative education services for students to acquire the digital competencies and skills throughout their educational trajectories. The main objectives of the law are the following: • Implementing a large-scale technological and digital transformation to accelerate learning among students through the definition of plans that address the multiple dimensions of the transformation, including the use of flexible modalities for the delivery of education services in all grades. • Providing the education system progressively with access to the necessary technological infrastructure, includ- ing energy, connectivity, digital devices, digital content, and education platforms to support pedagogy and accel- erate learning among students. • Consolidating and integrating the progress made during the pandemic for the development of education plat- forms and digital content available for public schools. • Promoting innovations and facilitating pre-service and in-service teacher training to guarantee their curricular pertinence and the mastery of the new teaching-learning models, in alignment with the technological and digital transformation of the education sector. • Implementing a national curriculum that is strengthened by the digital competencies needed for the develop- ment of 21st century skills, in all education levels across public and private schools • Creating a continuous evaluation system that promotes innovation, research, and technological development, fo- cused on improving decision making, the efficacy of education platforms, the improvement of learning outcomes, and the impact evaluation of the implementation of this education public policy. The Digital Equity Law has set a landmark regional example on how to build from the lessons learned during the pandemic with a focus on the long term and the sustainability of the education policies. This law will be critical to articulate efforts and build partnerships within the public sector and with the private sector to work together to make sure education technologies translate into more learning for students, particularly those from disadvantaged back- grounds. The law is expected to benefit 935,522 students from the formal and non-formal components of the regular and non-regular education subsystem in Panama from initial, pre-school, primary, lower secondary, and upper sec- ondary education. Source: Digital Equity Law, Official Gazette of the Republic of Panama, April 6, 2022. Recovering and accelerating: Recovering from learning losses, improving well-being, and addressing digital gaps Two Years After | Saving a Generation 131 largely discontinued. Some countries built on pre-exist- At the same time, it will also be important to address ing or introduced new centrally driven formative assess- psychosocial health and well-being. The crisis also gen- ments. These complementary assessment efforts need to erated important disruptions in psychosocial health and be urgently stepped-up everywhere to diagnose the real well-being with implications for readiness to learn and learning levels of students, including recent learning loss- skills development. Countries should put in place diag- es. Formative assessments are especially promising at this nostic assessments to provide timely data on the effects stage. At the same time, international, regional and/or of the pandemic on these aspects and design comprehen- national standardized assessments should continue to be sive strategies to address them, which should be evaluat- fostered to generate comparable data for benchmarking ed. These strategies need to include enhanced training of learning losses and inform educational policies and educa- and support for teachers who are often called to be first tion sector planning. Overall, what is needed is a strength- responders. ened evaluation culture where formative and summative assessments work together to ultimately improve learning. Finally, as countries move forward through the recov- ery and acceleration stages, addressing digital divides Learning recovery programs should be scaled-up. Coun- should also be a priority. Efforts to connect families and tries should step-up initiatives and programs to recover schools should continue, with focus on both the access from learning losses, building on lessons from pre-existing and quality of the Internet access. This needs to be com- and new programs. These programs should focus on foun- bined with proper equipment and devices and enhancing dational literacy and numeracy skills. Good monitoring digital skills of students and teachers. Digital competen- and evaluation systems and strengthening institutional cies frameworks and periodic assessments of digital skills capacities to implement at scale will be critical. Teachers, should be mainstreamed. Assuming basic digital gaps tutors, parents, and other implementers need to be trained and constraints to make use of the data and technology and supported on how to target instruction, by teaching at innovations are addressed, there is potential to build on the right level of students. Targeted instruction has strong data and technology to improve service delivery through potential to also accelerate learning in the longer run. Sup- strengthened information systems, platforms to enhance porting a prompt recovery in early childhood education targeted learning, or, yet, innovative and sustainable re- will be essential to tackle lack of readiness to learn. mote and hybrid learning models. 132 Two Years After | Saving a Generation PHOTO BY: © MACIEJ CZEKAJEWSKI / SHUTTERSTOCK.COM Chapter 6 Conclusion Conclusion Two Years After | Saving a Generation 133 T he impact of the pandemic on education in LAC was strong, and affected schooling, learning, and skills development. While the full effect of the pandemic on into a sufficiently broad-based commitment and re- sponse. Countries rushed to respond to the crisis with re- mote learning packages, policies to retain students in the schooling is not year clear, household surveys and admin- education system, and, to a lesser extent, measures and istrative data collected for this report show that students programs to start recovering from learning losses, show- aged 6-14 seem to have been the most affected, with a ing noticeable efforts to address the severe disruptions, consistent decrease in attendance rates. But everybody but much more needs to be done. Worryingly, although is at risk. High levels of disengagement during distance countries went out of their way to deploy massive stimulus schooling are expected to lead to future dropouts through packages in response to the health crisis, only limited re- learning gaps, especially for vulnerable groups. In fact, sources, with the caveat that some funding may have sup- learning already took a huge toll. Updated simulations ported cross-sectorial efforts and that more recent infor- show a large increase in the proportion of 3rd and 6th grad- mation is not yet available, were allocated to the education ers not able to adequately understand and interpret a text sector. And once again here, LAC was disproportionately af- of moderate length. Documented losses from a handful fected, which has constrained an adequate fully-fledged re- of LAC countries with robust data to date point to large sponse. While in Europe and North America, roughly 4% of significant drops in grade-appropriate reading and math the stimulus packages went to education and training, the scores, erasing decades of improvement. And this is likely corresponding value for LAC, Africa and Asia was roughly only the tip of the iceberg as more assessments become 1%. Indeed, there were large disparities by income level: available. Early grades and vulnerable populations seem while in high-income countries, on average, countries al- to be the groups most disproportionately affected. Lastly, located 3% to education; in low- and lower-middle-income the impact on psychosocial health and well-being of stu- countries, the same figure was less than 1%. dents and teachers was noticeable, as was the one on digi- tal capabilities, which were put to test during the pandem- Against this background, and expecting increasing ic, where the pre-existing digital divide ended up being a schooling and learning gaps moving forward, the edu- significant hurdle. Some came to refer to the segregation cation recovery agenda in LAC is huge. Efforts need to be trend engendered during the remote learning experience stepped-up. The report is clear in its mandate: “return to in the pandemic as: Zoom vs. Whatsapp schools240. schooling” and “recover and accelerate learning” should be the top priority strategies for the agenda moving for- This severe disruption to the schooling and learning ward and require a full package of actions, which will need process of LAC students did not yet translate, however, to be broadly supported. These strategies and actions are Figure 6.1. Allocation to education in total COVID-19 stimulus packages, by region (million USD and %), 2021 446,627 467,603 4.6 18,002 17,490 4.1 3.92 16,002 3.6 14,002 3.1 12,002 2.9 2.6 10,002 2.1 8,002 1.25 1.6 6,002 4,002 1.1 0.82 2,002 0.64 0.6 478 664 0.41 2,344 2 0.1 Africa Arab States Asia Asia & The Paci c Europe Latin America & Global & The Paci c & North America The Caribbean Total Estimated Funding in Education (Million USD) % of Education in Total Stimulus Packages (IMF) Source: UNESCO (2021) 240 See Romero et al. (2021). Conclusion 134 Two Years After | Saving a Generation consistent with the sizeable inequalities that pervade the assessments should continue to be fostered to generate region, with different countries being at different stages comparable data for benchmarking learning losses and in- of the recovery agenda. Still, recovering, accelerating, and form educational policies and education planning. improving should be a common denominator and a unify- ing framework to save the potentially devastating human Third, it is crucial to implement at scale adequate learn- capital loss for the “pandemic generation”. The goal is sim- ing recovery programs. Learning losses are expected to be ple: we need to avoid that the exogenous shock suffered by widespread, and learning heterogeneity to have increased, the 170 million students in LAC that went through a real requiring a scale-up of reforms, initiatives and programs “educational turmoil” end up becoming an educational to recover. Several remedial interventions for learning re- tragedy with a “lifetime human capital disability”. covery have now proven to be effective, so countries could build on those adjusting them as needed, all of which The first task ahead is the “return to schooling”, as we highlight the criticality of the new paradigm: targeted (as need to ensure the region is back on track to continue opposed to standardized) learning. All programs should reducing its OOSP. Prior to the pandemic, the region was focus on foundational literacy and numeracy skills. Good steadily reducing the number of children and adolescents monitoring and evaluation systems and strengthening in- of compulsory school age that were outside the education- stitutional capacities to implement at scale will be critical. al system. The pandemic seems to have paused this down- Teachers, tutors, parents, and other implementers need ward trend, at least for the 6-to-14-year-old group. Risks of to be trained and supported on how to target instruction, future dropouts are also very high. The proposal to redress by teaching at the right level of students. This approach is the setback is two-fold. First and foremost, it is essential critical to also improve learning in the longer run. to instill a sense of urgency to safely reopen all schools, es- pecially in those few places where educational institutions The two key strategies comprised within the school- continue to remain closed. This can be achieved notably ing and learning recovery agenda need to be accom- by consistently applying a set of safety standards that panied by a solid approach to address psychosocial could help along the process. The second element of the health. Monitoring mental health and well-being of stu- strategy is to reach out to re-enroll, hence the importance dents and teachers is just as critical to track education- of communication campaigns to reach those most likely al progress as standardized learning assessments are for to have decided to drop out, and prevent future dropouts, documenting the status of learning. Recovery programs through a combination of smart policies and programs, es- and interventions to support teachers and students’ psy- pecially targeting the most vulnerable groups. EWSs and chosocial health and well-being will be a vital element in EMISs must be essential ingredients of the strategy, as the the recovery. report describes. In addition, the pandemic surfaced the challenge of An even more urgent complementary strategy given the bridging the digital gap. Education systems must get emerging evidence on learning losses is the “recovery ready for any new eventual “sequel” of the pandemic, ir- and acceleration of learning”, capital to the skills build- respective of the form of the shock. The experience of the ing process that is fostered in classrooms. This strategy, forced (and rushed) application of remote/distance learn- nonetheless, requires prioritization, as the dimension of ing delivery channels has left an indelible mark: learning the task could look daunting otherwise. A three-tiered ap- effectiveness rested, for the most part, on the quality of proach is also offered by the report. First, to prioritize the access to digital infrastructure and the quality of digital curriculum. The report suggests to zero in on foundational skills. Building the backbone of a new “back-up system” skills, namely critical literacy – involving reading, writing, to replace in-person education – or simply to continue and speaking skills – and numeracy. Transferable skills providing hybrid education where this is still the case – (such as problem solving and critical thinking) should also is therefore critical. But this requires investments in both be integrated in the prioritized curriculum. Second is to physical and human ICT capital. These investments are measure the extent of learning losses. In-classroom as- also needed to spur innovations in service delivery. sessments, especially those of formative nature, are core tools to diagnose the depth, breadth, and characteristics LAC’s agenda for action is clear. Countries need to en- of the learning that was lost during the pandemic, both hance their investments to recover, accelerate and improve of the “forgotten” and the “foregone” types. At the same learning. The agenda for action developed in this report time, international, regional and/or national standardized has highlighted the urgency of four clear commitments to Conclusion Two Years After | Saving a Generation 135 Figure 6.2. The four commitments of the education recovery in LAC Learning and well-being: Schooling: Recover and enhance Teachers: Advocacy and nancing: Leave no one behind foundational skills and Value and support Place education recovery and prevent droputs well-being teachers at the top of the public agenda • Reopen all schools in a • Consolidate curricula with • Address human resource • Fund schools’ safety sustainable way. focus on foundational (and gaps and strengthen protocols, digital upgrades, • Support inclusive information transferrable) skills. teacher professional and all programs and campaigns on school • Assess learning levels, with development. measures at scale. re-enrollment, deploy cash summative and formative • Strengthen teachers' • Mobilize multiple transfer programs and ensure assessments. pedagogical and digital stakeholders in the delivery learning materials, informa- • Scale up initiatives and skills. and nancing e ort. tion, and services are programs to recover from • Support health and • Look for e ciency gains in accessible to all, especially learning losses, with focus well-being of teachers. use and allocation of those most vulnerable. on teaching at the level of resources, through better • Implement early warning the student and building on data, technological systems to identify and pre-existing and new innovations and institution- monitor students at risk of strategies and programs. al reforms for service dropping out. • Address psychosocial delivery improvements. • Address psychosocial health health and well-being of and well-being of students. students. • Address digital divides, with a • Address digital divides, focus on connectivity, digital with a focus on connectivi- skills and institutional ty, digital skills and capacity. institutional capacity. achieve this goal: (i) a commitment to schooling – so as to even lower than in the previous school year.241 The school- ensure that no student is left behind and/or drops-out; (ii) ing and learning recovery package has a “price-tag”. Im- a commitment to learning and well-being – to prioritize proving school safety has a cost and so implementing at foundational skills and foster appropriate levels of hu- scale programs to support attendance and recover from man capital formation along the way; (iii) a commitment learning losses. IAD (2021b) shows that $47 billion is re- to teachers – to make sure teachers are valued and sup- quired to connect the region to the Internet for education- ported throughout; and (iv) a commitment to advocacy and al purposes by 2030. Notwithstanding the primary role of financing – insofar as the education recovery agenda is a ministries of education and public funding, the response responsibility of all and needs sufficient, and wisely used, can also build on the partnerships that have developed resources for its implementation. The report has intended during the crisis. There have been multiple examples of to provide an actionable roadmap to ascertain which are such partnerships in the region that have been reviewed solid measures, programs, interventions, and strategies to in this report. To cite a couple of examples: ministries/sec- achieve the first three commitments and flag some risks retaries of education, assessment agencies, universities, and opportunities to achieve the fourth commitment. private foundations, NGOs, and/or civil society have been involved in the design and implementation of formative as- The fourth commitment is about advocacy and funding, sessments and learning recovery programs; collaborations which also presents opportunities for partnerships and between telecommunication agencies, Internet providers efficiency improvements. A broad-based commitment and education ministries were critical to provide low- and response are essential to avoid that a real education- cost Internet connections and expand access to platforms al turmoil ends up becoming an educational tragedy with during school closures. Further strengthening and devel- large economic and social implications. The region needs oping these partnerships for education is an opportunity to protect education budgets first. With rising inflation, af- provided by the crisis. There is also potential for efficiency fecting all income-level countries, real education budget improvements, to complement additional funding efforts, amounts for the current school year are expected to be building on data, technology, and institutional reforms. 241 UNICEF, UNESCO, and WBG (2022). References 136 Two Years After | Saving a Generation References Abufhele, A.; Bravo, D.; Lopez, F.; and Soto-Ramirez, P. (2021). Azevedo, J.P.; Akmal, M.; Cloutier, M.H.; and Wong, Y.N. “Developmental Losses in Young Children from Pre-Pri- (2022). “COVID-19 Learning Loss Simulations: Global mary Program Closures During the COVID-19 Pandem- Update”. Washington, D.C.: The World Bank Group. Un- ic.” Inter-American Development Bank. published Manuscript. Adelman, M. and Lemos, R. (2021). “Managing for Learning: Azevedo, J.P.; Hasan, A.; Goldemberg, D.; Iqbal S.; and Geven, Measuring and Strengthening Education Management K. (2020). “Simulating the Potential Impacts of COVID-19 in Latin America and the Caribbean”. Washington, DC: School Closures on Schooling and Learning Outcomes: A World Bank doi:10.1596/978-1-4648-1463-1. License: Set of Global Estimates”. Policy Research Working Paper Creative Commons Attribution CC BY 3.0 IGO. 9284; World Bank Group: Education Global Practice. Alexander, B.; Adams Becker, S.; and Cummins, M. (2016). Bandura, A. (1977). “Self-efficacy: Toward a Unifying Theo- “Digital Literacy: An NMC Horizon Project Strategic ry of Behavioral Change”. Psychological Review, 84(2), Brief”. Volume 3.3, October 2016. Austin, Texas: The 191-215. New Media Consortium. Banerji, R. and Chavan, M. (2016). “Improving Literacy and Angel-Urdinola, A. (2020). “Use of Adaptive Computer-Assist- Math Instruction at Scale in India’s Primary Schools: The ed Remediation Programs to Prevent Student Dropout in Case of Pratham’s Read India Program”. Journal of Edu- the Context of COVID-19”. World Bank Blogs. cational Change 17, no. 4, p. 453-475. Appelbaum, M. (2009). “The One-Stop Guide to Implement- Barron-Rodriguez, M., Cobo, C., Muñoz-Najar, A., and Sánchez ing RTI: Academic and Behavioral Interventions, K-12”. Ciarrusta, I. (2021a). “Remote Learning During the Glob- Corwin Press. al School Lockdown: Multi-Country Lessons”. Washing- Araujo, M. C.; Carneiro, P.; Cruz-Aguayo, Y.; and Schady, N. ton, D.C.: The World Bank Group. (2016). “Teacher Quality and Learning Outcomes in Barron-Rodriguez, M., Cobo, C., Muñoz-Najar, A., and Sánchez Kindergarten”.  The Quarterly Journal of Economics, Ciarrusta, I. (2021b). “What is Hybrid Learning? How Can 131(3):1415– 1453.  Countries Get it Right?”. World Bank Blog, April 27, 2021. Ardington, C.; Wills, G.; and Kotze, J. (2021). “COVID-19 Learn- Bashir, S. and Miyamoto. K. (2020). “Digital Skills: Frame- ing Losses: Early Grade Reading in South Africa”. Inter- works and Programs”. World Bank, Digital Economy for national Journal of Educational Development. Volume Africa (DE4A). 86, October 2021, 102480. Bassi, M.; Busso, M.; Urzúa, S.; and Vargas, J. (2012). “De- Arenas, A. and Gortázar, L. (2022). “Learning Loss One Year af- sconectados: Habilidades, Educación y Empleo en Améri- ter School Closures: Evidence from the Basque Country”. ca Latina” (Disconnected: Skills, Education and Employ- Working paper. EsadeEcPol-Center for Economic Policy. ment in Latin America). Washington, DC: Inter-American Arias, E.; Hincapié, D.; and Paredes, D. (2020). “Educar para Development Bank. la Vida. El Desarrollo de Habilidades Socioemocionales y el Bassi, M.; Busso, M.; and Muñoz, J. (2015). “Enrollment, Grad- Rol de los Docentes” (Educating for Life. Socio-emotional uation, and Dropout Rates in Latin America: Is the Glass Skills Development and the Role of Teachers). Washing- Half Empty or Half Full?”. Economía vol. 16, No.1 (Fall), ton D.C.: Inter-American Development Bank. pp. 113,115-156, Brookings Institution Press. Attanasio, O.; Cardona-Sosa, L.; Medina, C.; Meghir, C.; and Bellei, C.; Contreras, M.; Ponce, T.; Yañez, I.; Díaz, R.; and Viel- Posso, C. (2021). “Long Term Effects of Cash Transfer Pro- ma C. (2022). "The Fragility of the School-in-Pandemic grams in Colombia”. Cowles Foundation for Research in in Chile”. In: Reimers F.M. (ed.), Primary and Secondary Economics, Yale University. Discussion Paper No. 2293. Education During COVID-19. Springer, Cham. Azevedo, J.P. (2020). “Learning Poverty: Measures and Sim- Blyth, D.; Jones, S.; and Borowski, T. (2018). “SEL Frameworks ulations”. Policy Research Working Papers. The World - What Are They and Why Are They Important?”. Frame- Bank. doi:10.1596/1813-9450-9446 works Briefs Introductory Series No 1, Measuring SEL. References Two Years After | Saving a Generation 137 Bos, M.S.; Moffa, N.; Vegas, E.; and Zoido, P. (2017). “América Accumulating Effects of Individualized Literacy In- Latina y el Caribe en PISA 2015: ¿Sabemos Trabajar Juntos struction on Students’ Reading from First through Third para Resolver Problemas?” (Latin America and the Carib- Grade”. Psychological Science, 24, 1408–1419.  bean in PISA 2015: Do We Know How to Work Together Connor, C.M.; Phillips, B.M.; Kim, Y.; Lonigan, C.J.; Kaschak, to Solve Problems?). Note 11, CIMA, Inter-American De- M.P.; Crowe, E.; Dombek, J.; and Al Otaiba, S. (2018). velopment Bank. “Examining the Efficacy of Targeted Component Inter- Broadband Commission for Sustainable Development, In- ventions on Language and Literacy for Third and Fourth ternational Telecommunication Union, and UNESCO Graders Who Are at Risk of Comprehension Difficul- (2021). “Connecting Learning Spaces: Possibilities for ties”. Scientific Studies of Reading, 22(6), 462-484.  Hybrid Learning”, Working Group on Digital Learning. Cossi, J.; Dória, C.; Leal, O.; Lichand, G. (2021). “The Impacts Burns, B. and Luque, J. (2014). “Great Teachers: How to Raise of Remote Learning in Secondary Education: Evidence Student Learning in Latin America and the Caribbean”. from Brazil during the Pandemic”. Education Division. Washington DC: The World Bank. Technical Note N°IDB-TN-02214. Washington, DC: In- Busso, M.; Cristia, J.; Hincapié, D.; Messina, J.; and Ripani, L. ter-American Development Bank. (2017). “Aprender Mejor: Políticas Públicas para el Desar- Cossi, J.; Dória, C.; Leal, O.; and Lichand, G. (2022). "Reopen- rollo de Habilidades” (Learning Better: Public Policies for ing Schools in the Pandemic Did Not Increase COVID-19 Skills Development). Washington D.C.: Banco Interamer- Cases or Deaths in Sao Paulo State, Brazil”. Education icano de Desarrollo. Division. Technical Note N°IDB-TN-02415. Washington, Carbajal, F.; Tuzman D.; and Rovner H. (2022). “Cierre de Es- DC: Inter-American Development Bank. cuelas en Pandemia – Los Aprendizajes en Uruguay (School COVID-19 Excess Mortality Collaborators (2022). “Estimating Closures during the Pandemic – Learning Outcomes in Excess Mortality Due to the COVID-19 Pandemic: A Sys- Uruguay). The World Bank. Unpublished Manuscript. tematic Analysis of COVID-19-Related Mortality, 2020– Carneiro, P.; Crawford, C.; and Goodman, A. (2007). “The Im- 21”, The Lancet, Published online: March 10, 2022. pact of Early Cognitive and Non-Cognitive Skills on Later Crawford, M. and Oviedo M.E. (2022). “Early Grade Reading Outcomes”. Centre for the Economics of Education. Rainbow: A Quick Guide to Ending Learning Poverty (En- Case, A. and Deaton, A. (2017). “Mortality and Morbidity in glish)”. Washington, D.C.: World Bank Group. the 21st Century”. Brookings Papers on Economic Activi- Cunha, F.; Heckman, J.J.; Lochner, L.; and Masterov, D.V. ty 2017(1), 397-476. (2006). “Chapter 12. Interpreting the Evidence on Life CDC (2021). “Science Brief: Transmission of SARS-CoV-2 in Cycle Skill Formation”. In E. Hanushek and F. Welch, ed- K-12 Schools and Early Care and Education Programs – itors. Handbook of the Economics of Education, 1(12): Updated”. Centers for Disease Control and Prevention. 697–812, Amsterdam: North-Holland. Chakera, S.; Haffner, D.; and Harrop, E. (2020). “UNICEF East- Cunha, F. and Heckman, J.J. (2007). “The Technology of Skill For- ern and Southern Africa Region Working Paper – Struc- mation”. American Economic Review 97(2) May: 31–47. tured Pedagogy: For Real-Time Equitable Improvements Cunha, F. and Heckman, J. J. (2008). “Formulating, Identifying in Learning Outcomes”. UNICEF: Nairobi. and Estimating the Technology of Cognitive and Noncog- Chernyshenko, O.; Kankaras, M.; and Drasgow, F. (2018). nitive Skill Formation”. Journal of Human Resources, vol. “Social and Emotional Skills for Student Success and 43, no. 4, 2008, pp. 738–82. Well-Being: Conceptual Framework for the OECD Study Cunningham, W. and Villaseñor, P. (2016). "Employer Voices, on Social and Emotional Skills”. OECD Employer Demands, and Implications for Public Skills Chile’s Agency for the Quality of Education (2021). “Resulta- Development Policy Connecting the Labor and Educa- dos Diagnóstico Integral de Aprendizaje 2021” (Results tion Sectors”. Policy Research Working Paper No. 7582. from the Integral Learning Assessment 2021). Agencia Washington, D.C.: The World Bank. de Calidad de la Educación, Chile. Del Castillo, B. (2021). “Colegios sin Internet Llegan a 2.680 en Colombia’s National Education Ministry (2021). “Este Cuento es el País y Plan Busca Conectarlos de Aquí a 2022” (Schools Tuyo, una campaña para fomentar el acceso y la permanen- without Internet Reach 2,680 in the Country and Plan cia escolar en Colombia” (This Tale is Yours, a Campaign to Seeks to Connect Them by 2022). May 28. La Tercera Foster School Access and Retention in Colombia). News Newspaper. Note. Ministerio de Educación Nacional, Colombia. Delors, J; Amagi, I.; Carneiro, R.; Chung, F.; Geremek, B.; Gor- Connor, C. M.; Morrison, F. J.; Fishman, B.; Crowe, E. C.; Al ham, W; Kornhauser, A.; Manley, M.; Padrón, M.; Savané Otaiba, S.; and Schatschneider, C. (2013). “A Longi- M.; Singh,K.; Stavenhagen, R.; Won Suhr, M.; and Nan- tudinal Cluster-Randomized Controlled Study on the zhao, Z. (1996). “Learning: The Treasure Within”. Report References 138 Two Years After | Saving a Generation to UNESCO of the International Commission on Educa- Fukumoto, K.; McClean, C.; and Nakagawa, K. (2021). “No tion for the Twenty-first century. Paris: UNESCO. Causal Effect of School Closures in Japan on the Spread of Duckworth, A.L.; Peterson, C.; Matthews, M.; and Kelly, D. COVID-19 in Spring 2020”. Nature Medicine, Volume 27. (2007). “Grit: Perseverance and Passion for Long-Term Fundação Lemann (2021). “O Impacto da Pandemia na Al- Goals”. Journal of Personality and Social Psychology, fabetização no Brasil” (The Impact of the Pandemic on 92(6), 1087. Education in Brazil). Press News published at https:// Duckworth, A.L. and Seligman, M.E.P. (2005). “Self-Discipline fundacaolemann.org.br/noticias/o-impacto-da-pan- Outdoes IQ in Predicting Academic Performance of Ado- demia-na-alfabetizacao-no-brasil. lescents”. Psychological Science 16(12). 939–944. Galloway, T.; Lippman, L.; Burke, H.; Diener, O.; and Gates, S. Durlak, J.; Weissberg, R.; Dymnicki, A.; Taylor, R.; and Schell- (2017). “Measuring Soft Skills and Life Skills in Interna- inger, K. (2011). “The Impact of Enhancing Students’ So- tional Youth Development Programs: A Review and In- cial and Emotional Learning: A Meta-Analysis of School- ventory of Tools”. Washington, DC: USAID’s YouthPower Based Universal Interventions”. Child development Implementation IDIQ- Task Order 1, YouthPower Action. 82(1). 405-432. García, S.; Maldonado, D.; Llanes, L.M.; and Rodríguez, M. Elacqua, G.; Hincapié, D.; Vegas, E.; and Alfonso, M. (2018). (2021). “Desigualdad en la Educación en Colombia du- "Profesión: Profesor en América Latina, ¿Por qué se Per- rante el COVID-19” (Inequality in Education in Colombia dió el Prestigio Docente y Cómo Recuperarlo?” (Profes- during COVID-19). Universidad de los Andes and Fun- sion: Teacher in Latin America, Why Teacher’s Prestige dación Barco. Has Decayed and How to Regain It?). Washington, DC. García, S.; Maldonado, D.; Perry, G.; Rodríguez, C.; and Inter-American Development Bank. Saavedra, J. E. (2014). “Tras la Excelencia Docente: Cómo Ertem, Z.; Schechter-Perkins, E.; Oster, E.; Van den Berg, P.; Mejorar la Educación de Todos los Colombianos” (Seek- Epshtein, I.; Chaiyakunapruk, N.; Wilson, F.; Perencev- ing Teaching Excellency: How to Improve Education ich, E.; Pettey, W.; Branch-Elliman, W.; and Nelson, R. Quality for all Colombians). Fundación Compartir, Pun- (2021). “The Impact of School Opening Model on SARS- to Aparte.   CoV-2 Community Incidence and Mortality”. Nature Med- GEEAP (2020). “Cost-Effective Approaches to Improve Glob- icine, Volume 27. al Learning. What Does Recent Evidence Tell Us Are Estrada, R. and Lombardi, M. (2020). “Skills and Selection into “Smart Buys” for Improving Learning in Low- and Mid- Teaching: Evidence from Latin America”. Caracas: CAF. dle-Income Countries?”. Recommendations of the Glob- European Commission (2019). “The 2018 International Com- al Education Evidence Advisory Panel [K. Akyeampong, puter and Information Literacy Study (ICILS) – Main T. Andrabi, A. Banerjee, R. Banerji, S. Dynarski, R. Glen- Findings and Implications for Education Policies in Eu- nerster, S. Grantham-McGregor, K. Muralidharan, B. Pip- rope”. Luxembourg: Publications Office of the European er, J. Saavedra, S. Schmelkes, and H. Yoshikawa]. Wash- Union. ington D.C.: The World Bank Group; London: The Foreign, European Schoolnet (2021). “Recovery and Resilience Plans Commonwealth, and Development Office; and Building for Education: Agile Collection of Information”, Brussels, Evidence in Education (BE2). Belgium. GEEAP (2022). “Prioritizing Learning during COVID-19: The Faria, A.M.; Sorensen, N.; Heppen, J.; Bowdon, J.; Taylor S.; Eis- Most Effective Ways to Keep Children Learning during ner, R.; and Foster, S. (2017). “Getting Students on Track and Post-Pandemic”. Global Education Evidence Advi- for Graduation: Impacts of the Early Warning Interven- sory Panel [K. Akyeampong, T. Andrabi, A. Banerjee, R. tion and Monitoring System after One Year”. American Banerji, S. Dynarski, R. Glennerster, S. Grantham-Mc- Institutes for Research. Gregor, K. Muralidharan, B. Piper, S. Ruto, J. Saavedra, Feijao, C.; Flanagan, I.; Van Stolk, C.; and Gunashekar S. S. Schmelkes, and H. Yoshikawa]. Washington D.C., Lon- (2021). “The Global Digital Skills Gap: Current Trends don, Florence: The World Bank, FCDO, and UNICEF Office and Future Directions”. Santa Monica, CA: RAND Corpo- of Research - Innocenti. ration. Haimovich, F.; Vázquez, E.; and Adelman, M. (2021). "Scalable Friedman, H. S.; Kern, M. L.; and Reynolds, C. A. (2010). “Per- Early Warning Systems for School Dropout Prevention: sonality and Health, Subjective Well-Being, and Longev- Evidence from a 4.000-School Randomized Controlled ity”. Journal of Personality, 78(1), 179-216. Trial”. Policy Research Working Paper; No. 9685. World Fuchs, D. and Fuchs, L. S. (2006). “Introduction to Response Bank, Washington, DC: The World Bank Group. to Intervention: What, Why and How Valid is It?”. Read- Heckman, J.J. and Kautz, T. (2012). “Hard Evidence on Soft ing Research Quarterly. 41(1) 93–99. Skills”. Labour Economics 19(4). 451-464. References Two Years After | Saving a Generation 139 Heckman, J.J. and Kautz, T. (2013). “Fostering and Measuring IDB (2020). “Iniciativas de Alimentación Escolar durante la Skills: Interventions that Improve Character and Cog- Emergencia Sanitaria” (School Feeding Initiatives during nition”. National Bureau of Economic Research (NBER) the Health Emergency). Washington, DC: Inter-American Working Paper 19656. Cambridge, MA: National Bureau Development Bank. of Economic Research. IDB (2022). “How to Reboot Education Post-Pandemic – De- Heckman, J.J. and Rubinstein, Y. (2001). “The Importance of livering on the Promise of a Better Future for Youth”. Noncognitive skills: Lessons from the GED Testing Pro- Washington, DC: Inter-American Development Bank. gram”. American Economic Review 91(2). 145-149. ITA (2021). “Measuring the Impact of COVID-19 on Education Heckman, J. J.; Stixrud, J.; and Urzúa, S. (2006). “The Effects of in Pakistan”; Idara-e-Taleem-o-Aagahi (Centre of Educa- Cognitive and Noncognitive Abilities on Labour Market tion and Consciousness), Lahore: Pakistan. Outcomes and Social Behavior”. Journal of Labour Eco- IMF (2021). World Economic Outlook Database. nomics, 24(3): 411–482. INEEd (2021). “Aristas 2020. Primer Informe de Resultados de Herrera, L.; Buitrago, R.; Lorenzo, O.; and Badea, M. (2015). Tercero y Sexto de Educación Primaria” (Aristas 2020. “Socio-Emotional Intelligence in Colombian Children of First Report with Results for Third and Sixth Grades). In- Primary Education”. An Analysis in Rural and Urban Set- stituto Nacional de Evaluación Educativa, Uruguay. tings. Procedia-Social and Behavioral Sciences 203. 4-10. Ismail, S.; Saliba V.; Lopez Bernal, J.; Ramsay, M.E.; and Lad- Herrero-Tejada, A.; Flórez, A.; Stanton, S.; and Fiszbein, A. hani, S. (2021). “SARS-CoV-2 Infection and Transmission (2020). "Cambios e Innovación en la Práctica Docente in Educational Settings: A Prospective, Cross-Sectional en la Crisis del COVID-19” (Changes and Innovation in Analysis of Infection Clusters and Outbreaks in England”. the Teaching Practice during the COVID-19 Crisis). In- The Lancet, Vol.21, pp. 344-353. ter-American Dialogue and Teach for All. Jisc (2014). “Developing Digital Literacies”. Online Guide, Herrero-Tejada, A.; López, M.; Finoli, M.; Hevia, F.; Stanton, S.; available at https://www.jisc.ac.uk/guides/develop- and Fiszbein, A. (2022). "Evaluación en Pandemia: ¿Cómo ing-digital-literacies. Diagnosticamos las Pérdidas de Aprendizajes para Infor- J-PAL Abdul Latif Jameel Poverty Action Lab. (2019). “Tailor- mar los Esfuerzos de Recuperación?” (Pandemic Evalua- ing Instruction to Students’ Learning Levels to Increase tion: How to Diagnose Learning Losses to Inform Recov- Learning”. J-PAL Policy Insights. ery Efforts). Inter-American Dialogue. Jiménez, J. E.; Gutiérrez, N.; and de León, S. C. (2020). “Uni- Hevia, F. J.; Vergara-Lope, S.; Velásquez-Durán, A.; and versal Screening for Prevention of Reading, Writing, and Calderón, D. (2022). “Estimation of the Fundamental Math Disabilities in Spanish”. Journal of Visualized Ex- Learning Loss and Learning Poverty Related to COVID-19 periments: JoVE, (161), 10.3791/60217. Pandemic in Mexico.” International Journal of Educa- Jones, S.; Bailey, R.; Brush, K.; and Nelson, B. (2019). “Intro- tional Development. duction to the Taxonomy Project: Tools for Selecting and Hoskins, B. and Liu, L. (2019). “Measuring Life Skills in the Aligning SEL Frameworks”. Frameworks Briefs Compara- Context of Life Skills and Citizenship Education in the tive Series No 1, Measuring SEL. Middle East and North Africa”. United Nations Children’s Jones, S.; Bailey, R.; Brush, K.; Nelson, B.; and Barnes, S. Fund (UNICEF) and the World Bank. (2016). "What is the Same and What is Different? Mak- Hu, D.; Zhang, H.; Sun, Y.; and Li, Y. (2021). “The Effects of the ing Sense of the «Non-Cognitive» Domain: Helping Edu- Measures against COVID-19 Pandemic on Physical Activ- cators Translate Research into Practice”. Harvard Gradu- ity among School Aged Children and Adolescents (6–17 ate School of Education. years) in 2020: A Protocol for Systematic Review”. PLoS Kattan, R. and Székely M. (2017). "Analyzing Upper Second- ONE 16(7): e0255520. ary Education Dropout in Latin America through a Co- IAD (2021a). “El Uso de la Tecnología para Innovar la Prácti- hort Approach". Journal of Education and Learning. Vol.6, ca Docente: Retos y Lecciones Aprendidas en la Pandemia” No. 4. (The Use of Technology to Innovate in Teaching Practic- Kim, J.; Rose, P.; Tiruneh, D.T.; Sabates, R.; and Woldehanna, es: Challenges and Lessons Learned during the Pandem- T. (2021). Blog: “Learning Inequalities Widen Following ic). The Inter-American Dialogue. Washington, DC. COVID-19 School Closures in Ethiopia”. IAD (2021b). “El Estado de la Conectividad Educativa en Larraín, E. (2021). "Sistema de Alerta Temprana - Ministerio de América Latina - Desafíos y Oportunidades Estratégicas” Educación - Gobierno de Chile" (Early Warning System – (The State of Educational Connectivity in Latin Amer- Ministry of Education – Government of Chile). Presenta- ica – Challenges and Strategic Opportunities). The In- tion delivered at the “Recovering Schooling and Learn- ter-American Dialogue. Washington, DC. ing After COVID-19 in LAC Series #1 – Early Warning References 140 Two Years After | Saving a Generation Systems” Webinar organized by the Inter-American Di- Meinck, S.; Fraillon, J.; and Strietholt, R. (2022). “The Impact alogue, December 14. of the COVID-19 Pandemic on Education - International Lewis, K.; Kuhfeld, M.; Ruzek, E.; and McEachin, A. (2021). Evidence from the Responses to Educational Disruption “Learning during COVID-19: Reading and Math Achieve- Survey (REDS)”. UNESCO ment in the 2020-21 School Year”. Center for School and McCoy, D. C.; Cuartas, J.; Behrman, J.; Cappa, C.; Heymann, J.; Student Progress. Nwea Research. López Bóo, F.; Lu, C.; Raikes, A.; Richter, L.; Stein, A.; and Lippman, L.; Anderson Moore, K.; Guzman, L.; Ryberg, R.; Fink, G. (2021). “Global Estimates of the Implications of McIntosh, H.; Ramos, M.; Caal, S.; Carle, A.; and Kuhfeld, COVID-19-Related Preprimary School Closures for Chil- M. (2014). “Flourishing Children: Defining and Testing dren’s Instructional Access, Development, Learning, Indicators of Positive Development”. Springer Science and Economic Well-being”. Child Development, 92(5), and Business Media. e883-e899. Lopes, P. N; Brackett, M. A.; Nezlek, J. B.; Schütz, A.; Sellin, I.; MINEDUC-Chile (2020). “Deserción Escolar: Factores de Ries- and Salovey, P. (2004). “Emotional Intelligence and So- go y Prácticas de Prevención en Contexto de Randemia cial Interaction”. Personality and Social Psychology Bul- (School Drop-Out: Risk Factors and Prevention Practices letin, 30(8), 1018-34. in the Context of the Pandemic)”. Ministry of Education Loureiro, A.; Cruz, L.; Lautharte, I.; and Evans, D. (2020). "The of Chile, Research Unit. State of Ceará in Brazil is a Role Model for Reducing MINEDUC-Chile and WBG (2022). “Sistematización y Revisión Learning Poverty”. The World Bank. de la Experiencia de Chile para las Políticas y Medidas Ed- Loureiro, A.; Di Gropello, E.; and Arias, O. (2020). “There is ucativas en el Contexto de la Pandemia” (A Systematiza- No Magic: The formula for Brazil's Ceará and Sobral Suc- tion and Revision of the Chilean Experience with Edu- cess to Reduce Learning Poverty”. World Bank. Blogpost cation Measures and Policies in the Pandemic Context). Luna-Bazaldúa, D; Levin, V; and Liberman, J. (2020). “Guid- Mourshed, M.; Chijioke, C.; and Barber, M. (2010). “How the ance Note on Using Learning Assessment in the Pro- World’s Most Improved School Systems Keep Getting cess of School Reopening”. Washington, DC: The World Better”. New York, N.Y.: McKinsey and Company. Bank. Mulhern, C. (2020). “Beyond Teachers: Estimating Individu- Malala Fund (2020). “Girls’ Education and COVID-19. What al Guidance Counselors’ Effects on Educational Attain- Past Shocks Can Teach Us About Mitigating The Impact ment”. of Pandemics”. Malala Fund. Muñoz-Najar, A.; Gilberto, A.; Hasan, Amer; Cobo, C.; Azeve- Maldonado, J.E. and De Witte, K. (2021). “The Effect of School do, J, P.; and Akmal, M. (2021). “Remote Learning during Closures on Standardized Student Test Outcomes”. Brit- COVID-19: Lessons from Today, Principles for Tomor- ish Educational Research Journal. Vol 48, No 1, February row”. Washington, D.C.: World Bank Group. 2022, pp. 49-94 Näslund-Hadley, E.; Hernandez-Agramonte, J.M.; Montaño, K.; Malouff, J. M.; Thorsteinsson, E. B.; and Schutte, N. S. (2005). Namen, O.; Alpizar, G.; Luna, U.; Ochoa Foschini, L.; Ro- “The Relationship Between the Five-Factor Model of driguez, J.F.; Peña de Osorio, B.; Biehl, L.; Maragall, J.; Personality and Symptoms of Clinical Disorders: A Me- Mendez, C.; and Thomson, J. (2020). "Remote Initial Ed- ta-Analysis”. Journal of Psychopathology and Behavioral ucation and Mental Health during the COVID-19 Pan- Assessment, 27(2), 101-114. demic”. Washington, DC: Inter-American Development Mateo-Díaz, M. and Rodriguez-Chamussy, L. (2016). "Cash- Bank. ing in on Education: Women, Childcare, and Prosperity Nores, M.; Figueras-Daniel, A.; Lopez, M.A.; and Bernal, R. in Latin America and the Caribbean". Latin American De- (2018). “Implementing aeioTU: Quality Improvement velopment Forum. Alongside an Efficacy Study - Learning while Growing”. Mancebo, M.E. and Vaillant, D. (2022). “Programas de Recu- Annals of the New York Academy of Sciences: Special peración de Aprendizajes: Evaluación de la Evidencia y el Issue Implementation Research and Practice for Early Potencial para América Latina” (Learning Recovery Pro- Childhood Development. 1419 (2018) 201–217 grams: An Assessment of the Evidence and the Poten- Nugroho, D.; Youngkwang, J.; Akito, K.; and Lopez, F. (2021). tial for Latin America). Washington DC: Inter-American “It’s Not Too Late to Act on Early Learning: Understand- Dialogue. ing and Recovering from the Impact of Pre-Primary Ed- Martins, A.; Ramalho, N.; and Morin, E. (2010). “A Compre- ucation Closures during COVID-19”. Innocenti Research hensive Meta-Analysis of the Relationship between Briefs no. 2021-03. Emotional Intelligence and Health”. Personality and In- O’Connor, E.P and Witter Freeman, E. (2012). “District-Lev- dividual Differences, 49(6), 554-564. el Considerations in Supporting and Sustaining RTI References Two Years After | Saving a Generation 141 Implementation”. Psychology in the Schools, Vol. 49(3), Pratham. (2021). “Annual Status of Education Report”. Wiley Periodicals, Inc. Racine N.; McArthur B.A.; Cooke J.E.; Eirich R.; Zhu J.; and OECD (2015). “Skills for Social Progress: The Power of Social Madigan S. (2021). “Global Prevalence of Depres- and Emotional Skills”, OECD Skills Studies, OECD Pub- sive and Anxiety Symptoms in Children and Adoles- lishing, Paris cents During COVID-19:  A Meta-Analysis”.  JAMA Pedi- OECD (2016a). Country Profile: Argentina. atr. 2021;175(11):1142–1150. OECD (2016b). Country Profile: Mexico. Ramírez, M.J., and Valverde, G.A. (2021). “How to Ensure the OECD (2019). PISA 2018 Results. What School Life Means for Validity of National Learning Assessments? Priority Cri- Students’ Lives. Volume III. OECD Publishing, Paris. teria for Latin America and the Caribbean”. In: Manzi, J.; OECD (2020a). PISA 2018 Results. Effective Policies, Suc- García, M.R.; and Taut, S. (eds.). Validity of Educational cessful Schools. Volume V. OECD Publishing, Paris. Assessments in Chile and Latin America. Springer, Cham. OECD (2020b). “Learning Remotely When Schools Close. In- Rieble-Aubourg, S. and Viteri, A. (2020). “COVID-19: Are sights from PISA”. OECD Publishing, Paris. We Prepared for Online Learning?”. CIMA Brief, No. 20, OECD (2021). “Education in Brazil: An International Perspec- Washington, D.C., Inter-American Development Bank. tive”. OECD Publishing, Paris. Ripani, F. (2020). “Uruguay: Ceibal en Casa” (Ceibal at Home). Ontario Ministry of Education (2013). “Learning for All – A Education Continuity Stories Series, OECD Publishing, Guide to Effective Assessment and Instruction for All Paris. Students, Kindergarten to Grade 12”. [Ebook] (1st ed., Rodrigues Da Cruz Boari, L. and Crawford, M., (2022). “The pp. p. 11-24). Science of Reading in Practice: An Analysis of Instruc- Ospino, C. (2019). “El Rol de las Habilidades Socioemocionales tional Materials for Literacy in Selected States and Mu- para la Productividad en los Mercados Laborales del Siglo nicipalities in Brazil”. Washington, D.C: The World Bank XXI” (The Role of Socio-Emotional Skills on Labor Mar- Group. kets’ Productivity in the XXI Century). Technical Note Romero, C.; Krichesky, G.; and Zacarías, N. (2021).”Escuelas No. IDB-TN-1687. Washington, D.C.: Inter-American De- WhatsApp y Escuelas Zoom. Desigualdad y Segregación velopment Bank. Educativa durante la Pandemia COVID 19 en Argentina” Páez, J. and Castillo D. (2021). “Estrategia para el Acceso y la (WhatsApp Schools and Zoom Schools. Education In- Permanencia Educativa de Niños y Niñas” (Strategy for equality and Segregation during the COVID-19 Pandem- Boys and Girls’ Educational Access and Retention). To- ic in Argentina). Working Document N°10, Escuela de dos al Cole. Juntos Construimos Futuro. Gobierno Universidad Torcuato di Tella (UTDT). Perera, M., and Aboal, D. (2019). “The Impact of a Mathemat- Sánchez-Álvarez, N.; Extremera, N.; and Fernández-Berrocal, ics Computer-Assisted Learning Platform on Students’ P. (2016). “The Relation Between Emotional Intelligence Mathematics Test Scores”. Maastricht Economic and So- and Subjective Well-Being: A Meta-Analytic Investiga- cial Research institute on Innovation and Technology. tion”. The Journal of Positive Psychology, 11(3), 276-285. Perifanou, M.; Economides, A.; and Tzafilkou, K. (2021). São Paulo Education Secretariat (2021). "O Impacto da Pan- “Teachers’ Digital Skills Readiness during COVID-19 demia na Educação - Avaliação Amostral da Aprendiza- Pandemic”. International Journal of Emerging Technol- gem dos Estudantes (The Impact of the Pandemic on Ed- ogies in Learning (iJET) – eISSN: 1863-0383 – Vol. 16, ucation - Sampling Assessment of Student Learning). No. 08. Ministry of Education. Presentation available at Minis- Perusia, J.C. and Cardini, A. (2021). “Sistemas de Alerta Tem- try’s website. prana en la Educación Secundaria: Prevenir el Abandono Sartori, M. (2021); “Programa Brasil Na Escola – Ministerio de Escolar en la Era del COVID-19” (Early Warning Systems Educação - Brasil” (Brazil at School Program – Ministry for Secondary Education: Preventing School Drop-Out in of Education – Federal Government of Brazil). Presenta- COVID-19 Times); Centro de Implementación de Políti- tion delivered at the “Recovering Schooling and Learn- cas Públicas para la Equidad y el Crecimiento (Cen- ing After COVID-19 in LAC Series #1 – Early Warning ter for the Implementation of Public Policies for Equity Systems” Webinar organized by the Inter-American Di- and Growth - CIPPEC): Documento de Políticas Públicas alogue, December 14. (Public Policy Documents) #233, September 2021. Skovgaard Jensen, S. and Reimer, D. (2021). “The Effect Prada, M.F.; Rucci, G.; and Urzúa, S. (2019). “Training, Soft of COVID-19-Related School Closures on Students’ Skills and Productivity: Evidence from a Field Experiment Well-Being: Evidence from Danish Nationwide Panel in Retail”. IDB Working Paper Series No. IDB-WP-1015. Data”. SSM - Population Health. Volume 16, December Washington, D.C.: Inter-American Development Bank. 2021, 100945. References 142 Two Years After | Saving a Generation Stampini, M. and Tornarolli, L. (2012). “The Growth of Con- UNESCO-OREALC (2020). "¿Qué se Espera que Aprendan los ditional Cash Transfers in Latin America and the Carib- Estudiantes de América Latina y el Caribe? Análisis Curric- bean: Did They Go Too Far?”. Inter-American Develop- ular ERCE 2019” (What is Expected that Students from ment Bank. Latin America and the Caribbean Learn? ERCE 2019’s Stuart, E.A. and Dowdy, D. (2021).“Evidence-Based COVID-19 Curricular Analysis). Santiago, Chile: UNESCO-OREALC Policy-Making in Schools”. News and Views. (Regional Office for Latin America and the Caribbean). Tanaka, N.; Holla, A.; Pan, Y.; and Takeda, N. (2021). “Mea- UNESCO-OREALC (2021a). “Assessment of Socioemotional suring the Cost of School Re-Opening and Learning Re- Skills among Children and Teenagers of Latin America: covery”. Unpublished Manuscript. Washington, D.C.: The Framework for the ERCE 2019 Module Work.” Santiago, World Bank Group. Chile: UNESCO-OREALC (Regional Office for Latin Amer- Todos pela Educação (2021) “Nota Técnica: Taxas de Aten- ica and the Caribbean). dimento Escolar (Technical Note – School Attendance UNESCO-OREALC (2021b). “Los Aprendizajes Fundamentales Notes)”. Technical Note. en America Latina y el Caribe - Evaluación de Logros de los Tomasik, M.; Helbling, L.; and Moser, U. (2021). “Education- Estudiantes – ERCE 2019 – Resumen Ejecutivo” (Founda- al Gains of In-Person vs. Distance Learning in Primary tional Learning in Latin America and the Caribbean – As- and Secondary Schools: A Natural Experiment during the sessment of Learning Outcomes – ERCE 2019 – Execu- COVID-19 Pandemic School Closures in Switzerland” In- tive Summary). Santiago, Chile: Regional Office for Latin ternational Journal of Psychology, Vol.56 (4): pp.566-576. American and the Caribbean (OREALC). Trucco, D. and Palma, A. (2020). “Childhood and Adolescence UNESCO-OREALC (2021c). “Resultados Regionales: Habili- in the Digital Age. A Comparative Report of the Kids On- dades Socioemocionales en América Latina y el Caribe: Es- line surveys on Brazil, Chile, Costa Rica, and Uruguay”. tudio Regional Comparativo y Explicativo (ERCE 2019)” Project Documents (LC/TS.2020/18.REV.1), Santiago, (Regional Results: Socioemotional Skills in Latin Ameri- Economic Commission for Latin America and the Carib- ca and the Caribbean: ERCE 2019). Santiago, Chile: UNE- bean (ECLAC). SCO-OREALC (Regional Office for Latin America and the UNESCO (2019). “Recommendations on Assessment Tools for Caribbean). Monitoring Digital Literacy within UNESCO’s Digital Liter- UNESCO-OREALC and UNICEF-LACRO (2022). "Educación acy Global Framework”. UNESCO Institute for Statistics. en América Latina y el Caribe en el Segundo Año de la UNESCO (2020). “COVID-19 Education Response: How Many COVID-19” (Education in Latin America and the Carib- Students Are at Risk of Not Returning to School? Advo- bean in the Second Year of COVID-19”). Santiago, Chile: cacy Paper”. UNESCO Digital Library. UNESCO-OREALC (Regional Office for Latin America and UNESCO (2021); “Uneven Global Education Stimulus Risks the Caribbean). Widening Learning Disparities”. Program and Meeting UNICEF (2019a). “Global Framework on Transferable Skills”. Document. UNICEF, New York. UNESCO and UNICEF (2020). “In-Person Schooling and UNICEF (2019b). “Argentina: Every Child Learns. UNICEF Edu- COVID-19 Transmission: A Review of the Evidence”. cation Strategy 2019-2030”. UNICEF, Argentina. UNESCO, UNICEF, and WBG (2021c). “COVID-19 Learning UNICEF (2020). “Guidance on Distance Learning Modalities Losses: Rebuilding Quality Learning for All in the Mid- to Reach All Children and Youth During School Closures. dle East and North Africa". UNESCO, Paris, UNICEF, New Focusing on Low- and No-Tech Modalities to Reach the York, and World Bank, Washington, DC. © World Bank. Most Marginalized”. UNICEF, New York. UNESCO, UNICEF, WBG, and OECD (2021a). “What’s Next? UNICEF (2021a). “El Impacto de la Pandemia COVID-19 en las Lessons on Education Recovery: Findings from a Survey Familias con Niñas, Niños y Adolescentes - Quinta ronda of Ministries of Education amid the COVID-19 Pandem- - Informe de resultados” (The Impact of the COVID-19 ic”. Paris, New York, Washington D.C.: UNESCO, UNICEF, Pandemic on Households with Female and Male Chil- World Bank. dren and Adolescents - Fifth Round - Results Report). UNESCO, UNICEF, WBG, and OECD (2021b). Survey on Na- UNICEF (2021b). “COVID-19 and School Closures. One year of tional Education Responses to COVID-19 School Clo- Education Disruption”. sures, Round 3. Paris, New York, Washington D.C.: UNE- UNICEF (2021c) “Planea. Nueva Escuela para Adolescentes SCO, UNICEF, World Bank, OECD. (Planea. New School for Adolescents)”. Technical Note. UNESCO, UNICEF, WBG, World Food Programme, and UNHCR. UNICEF (2021d). “Reopening With Resilience: Lessons from (2021). “Framework for Reopening Schools Supplement: Remote Learning During COVID-19 – Latin America and From Reopening to Recovery - Key Resources”. the Caribbean”. Innocenti Research Report. References Two Years After | Saving a Generation 143 UNICEF (2021e). “The Importance of Transferable Skills De- Pandemic’s Impact on the Education Sector in Latin velopment in Latin America and the Caribbean”. Discus- America and the Caribbean”. World Bank, Washington, sion Paper. DC. UNICEF (2021f) “The State of the World’s Children 2021 – WBG (2021b). “COVID-19 in LAC: High Frequency Phone Sur- On My Mind: Promoting, Protecting and Caring for Chil- veys - Technical Note". World Bank, Washington, DC. © dren’s Mental Health” World Bank. UNICEF and ITU (2020). “How many Children and Young Peo- WBG (2021c). “Encouraging Re-Enrollment”. April 2021. ple Have Internet Access at Home? Estimating Digital Con- WBG (2021d). “Loud and Clear - Effective Language of Instruc- nectivity during the COVID-19 Pandemic.” International tion - Policies for Learning”. Washington, DC: The World Telecommunication Union, UNICEF, New York, 2020. Bank Group. UNICEF, UNESCO, and WBG (2022). “Where Are We on Educa- WBG (2022a). “Brazilian Human Capital Review”. Washington, tion Recovery?”. United Nations Children’s Fund (UNICEF). DC: The World Bank Group. Unpublished Manuscript. UNICEF LACRO (2020). “Report: Education on Hold - A Gener- WBG (2022b). “Educación para Todos Preservando la Salud: ation of Children in Latin America and the Caribbean are Una Mirada al Funcionamiento de la Educación Públi- Missing Out on Schooling because of COVID-19”. Pana- ca a Distancia en República Dominicana en Aiempos de ma: UNICEF’s Regional Office for the Latin America and COVID-19” (Education for All while Caring for Health: An the Caribbean (LACRO). Overview to Public Remote Learning Education in the Valverde, G.A.; Marshall, J.; and Soto, M.A. (2020). “El Apren- Dominican Republic during COVID-19). dizaje de Matemática en América Latina y el Caribe” (Math- WBG (2022c). “The Learning Crisis in Latin America and the ematics Learning in Latin America and the Caribbean). In Caribbean and the COVID-19 Pandemic: Sobering Re- Arias, A., Cristo, J., Cueto, S.: Aprender Matemática en el sults of a Deepening Trend”. Regional Note. Siglo XXI. A Sumar con Tecnología (Learning Mathematics WBG (forthcoming a). “Afro-Descendant Inclusion in Educa- in the XXI Century. Let’s Learn How to Add Using Technolo- tion. An Anti-Racist Agenda for Latin America”. Washing- gy). Washington DC: Inter-American Development Bank. ton, DC: The World Bank Group. Unpublished Manuscript. Valverde, G.A. and Ramírez, M.J. (2021). “Contemporary Prac- WBG (forthcoming b). “The State of Global Learning Poverty: tices in the Curricular Validation of National Learning 2022 Update”. Washington, DC: The World Bank Group. Assessments in Latin America: A Comparative Study of Unpublished Manuscript. Cases from Chile, Mexico, and Peru”. In: Manzi, J.; García, WBG, UNESCO and UNICEF (2021). The State of the Global M.R.; and Taut, S. (eds.) Validity of Educational Assess- Education Crisis: A Path to Recovery. Washington, D.C., ments in Chile and Latin America. Springer, Cham. Paris, New York: The World Bank Group, UNESCO and Van de Vijver, F. and Tanzer, N. K. (2004). “Bias and Equiva- UNICEF. lence in Cross-Cultural Assessment: An Overview”. Re- Zancajo, A.; Verger A.; and Bolea, P. (2022). “Digitalization vue Européenne de Psychologie Appliquée/European and Beyond: The Effects of COVID-19 on Post-Pandem- Review of Applied Psychology, 54, 119-135. ic Educational Policy and Delivery in Europe”. Policy and Van Laar, E.; Van Deursen, A.; Van Dijk, J.; and De Haan, Society, Volume 41, Issue 1. Pages 111–128. J. (2020). “Determinants of 21st-Century Skills and Zimmerman, K.; Brookhart, A.; Kalu, I.; Boutzoukas, A.; Mc- 21st-Century Digital Skills for Workers: A Systematic Lit- Gann, K.; Smith, M.; Maradiaga, G.; Armstrong, S.; We- erature Review”. SAGE Open. ber, D.; Moorthy, G.; and Benjamin, D. (2021). “Communi- Whizz Education (2021). “Measuring the impact of COVID-19 ty SARS-CoV-2 Surge and Within-School Transmission”. on learning in rural Kenya”. Pediatrics Volume 148, Issue 4. Wilichowski, T. and Cobo, C. (2020). “From Coping to Improv- Zimmerman, K.; Akinboyo, I.; Brookhart, M.; Boutzoukas, A.; ing and Accelerating: Supporting Teachers in the Pan- McGann, K.; Smith, M.; Maradiaga, G.; Armstrong, S.; Bri- demic and Beyond.” World Bank blog. stow, H; Parker, D.; Zadrozny, S; Weber, D.; and Benjamin, WBG (2021a). “Acting Now to Protect the Human Capital of D. (2021) Incidence and Secondary Transmission of SARS- Our Children: The Costs of and Response to COVID-19 CoV-2 Infections in Schools. Pediatrics. 2021;147(4). 144 Two Years After | Saving a Generation PHOTO BY: © NOWACZYK / SHUTTERSTOCK.COM Annexes Annex 1.1. Remote learning channels - strengths, limitations, and requirements Two Years After | Saving a Generation 145 Annex 1.1. Remote learning channels - strengths, limitations, and requirements Low-tech: Paper & printed material Strengths: Strengths: Requirements to be e ective: • Reach: Students without internet, • Limited content delivered & requires a • Frequent interactions with electricity and devices can certain level of reading skills and subject teachers & support from family members participate mastery to use e ectively. • Frequent assessment on content • Inclusion of the poor and most • No or little two-way interaction between • Communication between household & vulnerable teachers and students school • Student comprehension di cult to assess • E cient delivery of material & relevant • Student engagement & attendance di cult content developed to monitor   Low/medium-tech: Text messages & phone calls Strengths: • Reach: Students without internet Strengths: Requirements to be e ective: connection or computer can access • Limited learning material & content can be • Supplemental material for students materials on a frequent basis covered to continue to be engaged • Allows for interaction between • Little interactions & challenging to keep • Support from household members teachers and students track of multiple messages • Requires access to mobile phone device & • Potential to complement one-way • Di cult to assess learning credit communication modalities easily • Limited individual student access to mobile or phone Low/medium-tech: TV and Radio Strengths: Strengths: Requirements to be e ective: • Reach: Students without internet or • One-way communication • Radio and TV Devices computer access can participate • Limited tailoring of learning • Recording of classes & relevant content • Frequency, accessibility, and • Student engagement & attendance di cult • Frequent interactions with teachers & potential for large content coverage to monitor support from household members • Limited parental involvement • Student comprehension di cult to assess • Frequent assessments on content required for consumption • Risk excluding students without access to • Complementary material   Radio & TV High-tech: Online platforms Strengths: Strengths: Requirements to be e ective: • Wide range of functions to increase • Exclusion of students without devices • High tech devices & connectivity learning experience (adaptive • Connectivity challenges - access, costs and • Systems with platform license and high so ware, tailoring of learning, various e ectiveness level of management capacity contents, two/multi way communica- tion etc.) • In-person interaction missing • Teacher trained on platform • Provides learning exibility for • Teacher & student digital and socio-emo- students tional skills • Complement to virtual and face-to-face • Two/Multi-way communication solutions classroom learning • Monitoring of attendance, engagement and • Potential to facilitate monitoring of assessments of comprehension attendance, engagement, and learning Source: Barron-Rodriguez et al. (2021a); Muñoz-Najar et al. (2021); UNESCO (2020); UNICEF (2020); UNICEF (2021). Annex 1.2: School Closure Tracker - Methodology 146 Two Years After | Saving a Generation Annex 1.2: School Closure Tracker - Methodology  NESCO’s Global Monitoring of School Closures uses a 4-category classification for characterizing the closing of school U systems (Fully closed, Partially Closed, Fully Open, and Academic Break). This methodology builds on UNESCO’s breakdown by adding more variance to the definition of “partially closed” through the creation of 3 categories off the “Partially Closed” category, as follows: (i) a “Mostly Closed” category, which attempts to capture a situation where, on average, two-thirds or more (66%+) of the school system were closed on a given date; (ii) a “Partially Closed/Open” category, which comprise a so-called “average” situation where less than two-thirds, but more than one-third of the school system was closed (be- tween 33% and 66% closed); and (iii) a “Mostly Open” category, which tries to describe a situation where the proportion of schools closed in the system is lower than one-third (33%- Closed).  o carry out such a categorization, this report consolidates three sources of data: (i) UNESCO’s Global Monitoring of School T Closures (http://COVID19.uis.unesco.org/global-monitoring-school-closures-COVID19/); (ii) UNICEF’s Monthly Monitoring of School Closures (https://www.unicef.org/lac/respuesta-educativa-de-alc-frente-al-COVID-19); and (iii) country-specific web research (especially for dealing with data gaps, inconsistencies, school calendars, academic breaks, and/or simply to calcu- late shares of “partially closed”, as per the methodology described below).  The methodology for data handling, including a detailed step-by-step description is included below: Step 1: Create Original Tracker with UNESCO’s Database (4-Category Classification) •  Download UNESCO’s database (“Global monitoring of school closures/ Data”) (https://en.unesco.org/COVID19/ educationresponse#durationschoolclosures). •  Select desired fields (Date, Country, Region, Status) and delete the rest •  Filter LAC countries (45 total)242 Filter LAC countries (45 total) Anguilla Chile Haiti Saint Vincent and the Grenadines Antigua and Barbuda (OECS) Colombia Honduras Sint Maarten (Dutch part) Argentina Costa Rica Jamaica St. Martin (French part) Aruba Cuba Mexico Suriname Bahamas Curaçao Montserrat Trinidad and Tobago Barbados Dominica Nicaragua Turks and Caicos Islands Belize Dominican Republic Panama Uruguay Bermudas Ecuador Paraguay Venezuela Bolivia El Salvador Peru Virgin Islands (US) Brazil Grenada Puerto Rico British Virgin Islands Guatemala Saint Kitts and Nevis Cayman Islands Guyana Saint Lucia (OECS) 242 The UNESCO database does not include Puerto Rico, St Martin (French Part), or the Virgin Islands. These countries are, however, listed along with others be- low for convenience and completion. Annex 1.2: School Closure Tracker - Methodology Two Years After | Saving a Generation 147 •  Fix errors in dates. For example: errors in numeric dates (e.g. data downloaded in Feb 2022 for Uruguay, Saint Vincent and the Grenadines, Venezuela, British Virgin Islands). •  Filter out weekends. Assign numbers 1-7 for each weekday. Eliminate Saturday and Sundays. You should then have daily data per country. •  Assign a letter to each status (“Fully open”, “Partially open”, “Closed due to COVID-19”, and “Academic break”) •  Organize daily data in rows (1 row per country) for the period above. Place dates in columns and transpose the daily status for each country. Step 2: Disaggregate “Partially Closed” Into Three Categories Using Unicef Monthly Monitoring Reports •  Download UNICEF reports at https://www.unicef.org/lac/respuesta-educativa-de-alc-frente-al-COVID-19. •  Analyze the country-specific summaries, one by one, identifying the periods when schools were “partially closed”. Next, check in each report the school opening status attempt to re-classify the “partially closed” periods for the two extreme cases, i.e. •  Mostly Closed: If less than 33% of schools were open •  Mostly Open: if more than 66% of schools were open •  For the purposes of the new categorization, attempt to retrieve information on the share of total schools in the system that are closed. In the absence of data on share of total schools, use the share of students not attending face-to-face classes as a “proxy”. •  Record the information for each period (in a separate file) and then update the tracker. Step 3: Fill in Gaps With Further Research  •  If a given country is not included in UNICEF reports, or if information on school closures – or number of students in face-to-face classes - is not reported there either, perform a country-specific web research, prioritizing official informa- tion from Ministries of Education’s websites. In the absence of official information, use secondary sources, like most important digital newspapers of the given country. •  Update tracker. Annex 1.2: School Closure Tracker - Methodology 148 Two Years After | Saving a Generation STEP 4: Calculate the “share of effective school days” per category •  Add all the days per category. •  Estimate the “effective school days” indicator using the total number of days minus the total days on academic break. •  Estimate the “share of effective school days” per category (see example below). % of effective days Fully Mostly Partially Mostly Fully Effective Fully Mostly Partially Mostly Fully Closed Closed Open/ Open Open Days Closed Closed Open/ Open Open Closed (without Closed academic break) Honduras 293 85 16 0 8 402 73 21 4 0 2 Venezuela 304 0 27 16 10 357 85 0 8 4 3 El Salvador 228 160 0 0 7 395 58 41 0 0 2 Panama 276 70 51 0 8 405 68 17 13 0 2 Mexico 268 25 56 0 15 364 74 7 15 0 4 Ecuador 200 185 0 0 9 394 51 47 0 0 2 Peru 172 192 0 0 10 374 46 51 0 0 3 Jamaica 139 138 19 0 9 305 46 45 6 0 3 Guatemala 165 158 63 0 10 396 42 40 16 0 3 Annex 2.1. Compulsory Education Age for 45 countries in LAC Two Years After | Saving a Generation 149 Annex 2.1. Compulsory Education Age for 45 countries in LAC Compulsory Education Age for 45 countries in LAC 3 to 16 3 to 17 3 to 18 4 to 11 4 to 15 4 to 16 4 to 17 Peru Mexico Venezuela Barbados Guatemala Costa Rica Bolivia (Bolivarian (Plurinational Saint Martin Republic of) Turks and Caicos State of) (French part) Islands Brazil Honduras Uruguay 4 to 18 5 to 8 5 to 11 5 to 12 5 to 14 5 to 16 5 to 17 Argentina Puerto Rico Nicaragua Suriname Bahamas Bermuda Anguilla Curaçao Belize British Virgin Antigua and Islands Barbuda Sint Maarten Colombia (Dutch part) Dominica Cayman Islands Panama Grenada Dominican Republic Guyana Ecuador Montserrat Paraguay Saint Kitts and Nevis United States Virgin Islands Saint Lucia Saint Vincent and the Grenadines 6 to 11 6 to 12 6 a 14 6 to 16 6 to 18 Haiti Aruba El Salvador Trinidad and Chile Tobago Jamaica Cuba Note: As there is not available data for Saint Martin (French part), the French education system compulsory education age is reported in the table. Source: UNESCO, National Government data for each country. Annex 5.1. Flagship “remedial programs”: Characteristics and results 150 Two Years After | Saving a Generation Annex 5.1. Flagship “remedial programs”: Characteristics and results Name of the Country Characteristics Promoter and Target population and Evidence program implementation coverage agency Catch up Turkey An integral program Turkish National Children and youth Study based on papers Education with students in Ministry of Education between 10 and 14 years. and interviews, with Programme groups, and 8 weeks of and UNICEF positive results on the duration. Between years program’s participants. 2009-2010, the program covered Catch-up Programs 19,990 children in 61 provinces. Early Learners Members of the Catch up programs to USAID and OECS Pre-school and school Impact evaluation Programme Organisation improve language and children until 8-9 based on control of Eastern and lecture comprehension years. groups with results Caribbean suggesting a 50% States (OECS) Between years 2015- increase in reading. 2020, the program covered 73,000 children. Second Ethiopia Program until 4th Geneva Global Luminos Children and youth Impact Evaluation, with Chance grade during 8 hours Fund between 8 and 14 years. positive evidence with with an emphasis on respect to student’s reading, writing, and Society alliances Between years 2020- transition to regular mathematics, in small 2021 the program school. groups. covered 9,750 children and youth in the city of Acceleration Addis Ababa. Acelera Brasil Brazil Program based on Ayrton Senna Institute Children in 3rd to 5th Results evaluation, assessment diagnosis grades in primary. with some indicators and in grouped Ministry of Education such as the number of students according to During years 2000- read books, schooling their learning level 2020 the program attendance, and covered more than 1 cooperation between million children. teachers. Llegim in Spain A program designed Education Sciences Children between 8 to Impact evaluation parella to improve reading Institute in Universidad 11 years with ex-ante and ex- comprehension based Autónoma de Barcelona post data from the on tutoring between In 2011 the program intervention, suggesting students and family covered 303 children. positive results for all support. students. Tutoring Aula Global Colombia A program designed to National Education Children from 2nd to Impact evaluation improve reading and Ministry 5th grades. with ex-ante and mathematics, through ex-post data from semi-personalized Carvajal Foundation In the year 2020 the the intervention, tutoring. program covered suggesting positive 3,240 children in results for all 36 educational students. institutions Annex 5.1. Flagship “remedial programs”: Characteristics and results Two Years After | Saving a Generation 151 Name of the Country Characteristics Promoter and Target population and Evidence program implementation coverage agency TarL India Program based on Pratham ONG Children in 3rd to 5th Impact evaluation developing basic grades with positive results in mathematics and reading. reading skills, by Between years 2016- grouping students 2017 the program Teaching at the Right Level according to their covered 200,000 learning level, tutoring, children. and follow-up. TarL Mexico Ludic- and community- MIA-CIESAS Children and youth Impact evaluation based summer classes, Universidad between 3 to 14 years. with positive and designed to improve Veracruzana statistically significant mathematics and Between years 2016- effects on learning reading, by grouping 2020 the program for children who students according to covered 12,760 participated in the their learning level. children and youth in program. Veracruz. Vacances France School day extended National Education Children and youth The evaluation shows apprenantes program designed Ministry between 3 to 17 years. positive outcomes to improve learning, in learning among In the year 2019 the Extended Pedagogical Schedule developed in summer participants. camps. program covered 60,000 children. Jornada Dominican Educational program Ministry of Education Children and youth The program Extendida Republic of 8 hours daily from pre-primary, evaluation does not designed to improve primary, and show improvement in learning in different secondary. learning, but it does disciplines that are show improvement in part of the curriculum In years 2017-2018 socio-emotional skills. with school-feeding the program covered and health services. 1,162,849 children and youth. Mindspark India A program based on a Mindspark Children and youth Impact evaluation software that diagnoses in grades 1st to shows positive results wrong concepts and 10th (primary and in language and provides individualized secondary). mathematics. Computer-assisted learning content to help children in learning. It also In 2018, the program supports teachers. covered 400,000 children. Plataforma Uruguay A program based Centro Ceibal Children and youth in A longitudinal study Adaptativa de on individualized primary and secondary presented positive Matemática learning according to results in learning (PAM) the skill level of each Between years 2013- in mathematics for student. 2016 PAM was used primary students. by half the children in 3rd to 6th grade in primary education. Source: Mancebo and Vaillant (2022). Annex 5.2 152 Two Years After | Saving a Generation Annex 5.2 Box A.5.2. Digital competencies frameworks and assessments: A priority for post-pandemic times A relevant lesson learned during the pandemic is related to the digital competencies’ frameworks, crucial for assess- ing the abilities needed for an effective use of education digital platforms, for safe interactions in the digital world, and for selecting, curating, and producing digital contents for educational purposes. The Working Group on Technol- ogy and Innovation in Education, led by the Inter-American Dialogue, identified two main priorities in this area: (1), working for the consensus on a single, regional framework on competencies, on the basis of the current ones, to de- fine what digital skills are needed by professional teachers, and train them accordingly; (2), developing a in-service teacher training plan, prioritizing strategies on the basis of the common regional competencies framework, including self-evaluation tools for teachers to identify their most relevant needs, tools for the creation of peer learning commu- nities, and strategies for the certification of digital abilities. There are several examples and models to follow for the elaboration of digital competencies framework. In Latin America, there are some good practices. Chile has developed, since 2011, the Competencias para la Profesión Docente (Competencies for the Teaching Profession), which defines the digital skills expected for teachers at the national lev- el. It organizes the competencies in five dimensions (Pedagogical, Technical, Management, Social/Ethical/Legal, and Professional Development and Accountability). For each dimension, the framework includes specific competencies, descriptors associated to each competency, the required knowledge for the tasks included in the descriptors, and fields where the described knowledge applies. Colombia, in 2013, published the Competencias TIC para el Desarrollo Profesional Docente (ICT Competencies for Teacher Professional Development), that define competencies in five di- mensions (technological, pedagogical, communicational, managerial and research), establishing for each dimension the performance profiles: explorer, integrator, and innovator. Beyond the Latin American region, a comprehensive review of digital skills framework has been conducted by World Bank (Bashir and Miyamoto, 2020). The European Union DigComp identifies five areas (information and data literacy; communication and collaboration; digital content creation; safety; problem solving) and 21 competencies. For each competence, 8 proficiency levels are outlines, grouped in turn under four profiles: Foundational, Intermediate, Ad- vanced and Highly Specialized. UNESCO’s Institute of Statistics adapted this framework in 2018, adding device and career related competences. (Bashir and Miyamoto, 2020). UNESCO has also elaborated a set of recommendations on assessment tools for measuring digital skills, reviewing the advantages and disadvantages of different types of evaluation strategies: performance type assessments (in which individuals are monitored while performing digital tasks and solving problems in digital context), knowledge bases assessments (in which individuals respond test items) and self-assessments, where individual auto-administer questionnaires designed to self-diagnose areas for strengthening (UNESCO, 2019) Other relevant efforts, particularly relevant to the education field, are the Digital Competence Framework for Ed- ucators (DigCompEdu), that describes a concept of “digitally competent teacher” (https://ec.europa.eu/jrc/en/dig- compedu); the ISTE Standards for Educators, promoted by the International Society for Technology in Education to guide students in the development of digital skills (https://www.iste.org/standards/for-educators); and ICILS, the International Computer and Information Literacy Study, designed to measure “students’ ability to use computers to investigate, create, and communicate in order to participate effectively at home, at school, and in their communities (https://www.iea.nl/icils). Two Years After | Saving a Generation FOTOGRAFÍA DE: © THOMAS MICHAEL PERRY/THE WORLD BANK PHOTO BY: © GONZALO BELL / SHUTTERSTOCK.COM The COVID-19 pandemic caused the LAC region to suffer a triple curse. It was the region with the largest number of COVID-19-related deaths per million people. It was also the region with the largest economic downturn and with the worst medium-term economic prospects. Last but not least, it was among the top regions with the longest spells of school closures to date. The pandemic has had a sizeable impact on learning outcomes and skills development, eroding the already weak foundations of LAC’s education systems. It is estimated that the average student in LAC lost, to date, between 1 and 1.8 years of schooling, adjusted by learning. Average primary education scores in reading and math could could have fallen to levels of more than 10 years ago. The main policy recommendation is simple: recovery must focus on two essential strategies, returning to schooling and recovering from learning losses. This dual agenda should focus on four key commitments: (a) placing the education recovery at the top of the public agenda; (b) reintegrating all the children that abandoned school and ensure they stay in it; (c) recovering lost learning and ensure the socio-emotional well-being of children; and (d) valuing, supporting, and training teachers. IN COLLABORATION WITH