OVERVIEW OVERVIEW This booklet contains the overview, as well as a list of contents, from World Development Report 2018: Learning to Realize Education’s Promise, doi: 10.1596/978-1-4648-1096-1. A PDF of the final, full- length book, once published, will be available at https://openknowledge.worldbank.org/, and print copies can be ordered at http://Amazon.com. Please use the final version of the book for citation, reproduction, and adaptation purposes. © 2018 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: World Bank. 2018. “World Development Report 2018: Learning to Realize Education’s Promise.” Overview booklet. World Bank, Washington, DC. License: Creative Commons Attribution CC BY 3.0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party- owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. Cover design: Kurt Niedermeier, Niedermeier Design, Seattle, Washington. Interior design: George Kokkinidis, Design Language, Brooklyn, New York, and Kurt Niedermeier, Niedermeier Design, Seattle, Washington. Contents v Foreword vii Acknowledgments 1 Overview: Learning to realize education’s promise 4 The three dimensions of the learning crisis Learning outcomes are poor: Low levels, high inequality, slow progress Schools are failing learners Systems are failing schools Still, there are reasons for hope 16 How to realize education’s promise: Three policy responses Assess learning—to make it a serious goal Act on evidence—to make schools work for all learners Align actors—to make the whole system work for learning 27 Learning to realize education’s promise 37 Contents of the World Development Report 2018 iii Foreword Education and learning raise aspirations, set values, and ultimately enrich lives. The coun- try where I was born, the Republic of Korea, is a good example of how education can play these important roles. After the Korean War, the population was largely illiterate and deeply impoverished. The World Bank said that, without constant foreign aid, Korea would find it difficult to provide its people with more than the bare necessities of life. The World Bank considered even the lowest interest rate loans to the country too risky. Korea understood that education was the best way to pull itself out of economic misery, so it focused on overhauling schools and committed itself to educating every child—and educating them well. Coupled with smart, innovative government policies and a vibrant private sector, the focus on education paid off. Today, not only has Korea achieved universal literacy, but its students also perform at the highest levels in international learning assess- ments. It’s a high-income country and a model of successful economic development. Korea is a particularly striking example, but we can see the salutary effects of education in many countries. Delivered well, education—and the human capital it creates—has many benefits for economies, and for societies as a whole. For individuals, education promotes employment, earnings, and health. It raises pride and opens new horizons. For societies, it drives long-term economic growth, reduces poverty, spurs innovation, strengthens institu- tions, and fosters social cohesion. In short, education powerfully advances the World Bank Group’s twin strategic goals: ending extreme poverty and boosting shared prosperity. Given that today’s students will be tomorrow’s citizens, leaders, workers, and parents, a good education is an investment with enduring benefits. But providing education is not enough. What is important, and what generates a real return on investment, is learning and acquiring skills. This is what truly builds human capital. As this year’s World Development Report documents, in many countries and commu- nities learning isn’t happening. Schooling without learning is a terrible waste of precious resources and of human potential. Worse, it is an injustice. Without learning, students will be locked into lives of poverty and exclusion, and the children whom societies fail the most are those most in need of a good education to succeed in life. Learning conditions are almost always much worse for the disadvantaged, and so are learning outcomes. Moreover, far too many children still aren’t even attending school. This is a moral and economic crisis that must be addressed immediately. This year’s Report provides a path to address this economic and moral failure. The detailed analysis in this Report shows that these problems are driven not only by service delivery failings in schools but also by deeper systemic problems. The human capital lost v because of these shortcomings threatens development and jeopardizes the future of peo- ple and their societies. At the same time, rapid technological change raises the stakes: to compete in the economy of the future, workers need strong basic skills and foundations for adaptability, creativity, and lifelong learning. To realize education’s promise, we need to prioritize learning, not just schooling. This Report argues that achieving learning for all will require three complementary strategies: • First, assess learning to make it a serious goal. Information itself creates incentives for reform, but many countries lack the right metrics to measure learning. • Second, act on evidence to make schools work for learning. Great schools build strong teacher-learner relationships in classrooms. As brain science has advanced and educators have innovated, the knowledge of how students learn most effec- tively has greatly expanded. But the way many countries, communities, and schools approach education often differs greatly from the most promising, evidence-based approaches. • Third, align actors to make the entire system work for learning. Innovation in classrooms won’t have much impact if technical and political barriers at the sys- tem level prevent a focus on learning at the school level. This is the case in many countries stuck in low-learning traps; extricating them requires focused attention on the deeper causes. The World Bank Group is already incorporating the key findings of this Report into our operations. We will continue to seek new ways to scale up our commitment to education and apply our knowledge to serve those children whose untapped potential is wasted. For example, we are developing more useful measures of learning and its determinants. We are ensuring that evidence guides operational practice to improve learning in areas such as early-years interventions, teacher training, and educational technology. We are making sure that our project analysis and strategic country diagnoses take into account the full range of system-level opportunities and limitations—including political constraints. And we will continue to emphasize operational approaches that allow greater innovation and agility. Underlying these efforts is the World Bank Group’s commitment to ensuring that all of the world’s students have the opportunity to learn. Realizing education’s promise means giving them the chance not only to compete in tomorrow’s economy, but also to improve their communities, build stronger countries, and move closer to a world that is finally free of poverty. Jim Yong Kim President The World Bank Group vi | Foreword Acknowledgments This year’s World Development Report (WDR) was prepared by a team led by Deon Filmer and Halsey Rogers. The core team was composed of Samer Al-Samarrai, Magdalena Ben- dini, Tara Béteille, David Evans, Märt Kivine, Shwetlena Sabarwal, and Alexandria Valerio, together with research analysts Malek Abu-Jawdeh, Bradley Larson, Unika Shrestha, and Fei Yuan. Rafael de Hoyos and Sophie Naudeau were members of the extended team. Stephen Commins provided consultations support. Mary Breeding, Ji Liu, Christian Ponce de Leon, Carla Cristina Solis Uehara, Alies Van Geldermalsen, and Paula Villaseñor served as consul- tants. The production and logistics team for the Report consisted of Brónagh Murphy and Jason Victor. The Report is sponsored by the Development Economics Vice Presidency. Overall guid- ance for preparation of the Report was provided by Paul Romer, Senior Vice President and Chief Economist, and Ana Revenga, Deputy Chief Economist. In the early months of the Report’s preparation, guidance was provided by Kaushik Basu, former Senior Vice President and Chief Economist, and Indermit Gill, former Director for Development Policy. The team is also grateful for comments and guidance from Shantayanan Devarajan, Senior Director for Development Economics. The Education Global Practice and the Human Development Global Practice Group provided consistent support to the Report team. The team is espe- cially grateful for support and guidance provided by Jaime Saavedra, Senior Director, and Luis Benveniste, Director, of the Education Global Practice. The team received guidance from an advisory panel composed of Gordon Brown (who, together with the Chief Economist, cochaired the panel), Michelle Bachelet, Rukmini Banerji, Julia Gillard, Eric Hanushek, Olli-Pekka Heinonen, Ju-Ho Lee, and Serigne Mbaye Thiam. Although the team valued their advice and found it very useful, the views expressed in the Report do not necessarily reflect those of the panel members. The team also benefited at an early stage from consultations on emerging themes with the Chief Economist’s Council of Eminent Persons. Council members providing comments were Montek Singh Ahluwalia, François Bourguignon, Heba Handoussa, Justin Yifu Lin, Ory Okolloh, Pepi Patrón, Amartya Sen, Joseph Stiglitz, Finn Tarp, and Maria Hermínia Tavares de Almeida. Paul Holtz was the principal editor of the Report. Bruce Ross-Larson provided editorial guidance, and Sabra Ledent and Gwenda Larsen copyedited and proofread the Report. Kurt Niedermeier was the principal graphic designer. Alejandra Bustamante and Surekha Mohan provided resource management support for the team. Phillip Hay, Mikael Reventar, Anushka Thewarapperuma, and Roula Yazigi, together with Patricia da Camara and Kavita Watsa, provided guidance and support on communication and dissemination. Special thanks are extended to Mary Fisk, Patricia Katayama, Stephen Pazdan, and the World Bank’s Formal Publishing Program. The team would also like to thank Maria Alyanak, Laverne Cook, Maria del Camino Hurtado, Chorching Goh, Vivian Hon, Elena Chi-Lin Lee, Nancy Tee Lim, David Rosenblatt, and Bintao Wang for their coordinating roles. vii The team is grateful for generous support for preparation of the Report provided by the Knowledge for Change Program (KCP, a multidonor Trust Fund) and especially from the governments and development agencies of the following KCP donor countries: Finland, France, and Norway. Background and related research, along with dissemination, are being generously supported by the Bill & Melinda Gates Foundation, Early Learning Partnership Trust Fund, LEGO Foundation, and Nordic Trust Fund. Consultation events attended by government officials, researchers, and civil society organizations were held in Bolivia, Brazil, Canada, China, Côte d’Ivoire, Finland, France, Germany, India, Indonesia, Japan, Kenya, Malaysia, Mexico, Senegal, South Africa, Tanzania, Thailand, Turkey, the United Kingdom, and the United States, with participants drawn from many more countries. The team thanks those who took part in these events for their helpful comments and suggestions. Further information on these events is available at http://www .worldbank.org/wdr2018. Interagency consultations were held with the Association for the Development of Education in Africa (ADEA), Global Development Network (GDN), Global Partnership for Education (GPE), International Commission on Financing Global Education Opportunity (Education Commission), International Monetary Fund (IMF), Organisation for Economic Co-operation and Development (OECD), United Nations Children’s Fund (UNICEF), and United Nations Educational, Scientific, and Cultural Organization (UNESCO). Consulta- tions with bilateral development partners included representatives of the governments of Canada, Finland, Japan, the Republic of Korea, Norway, and Sweden, and of Australia’s Department of Foreign Affairs and Trade (DFAT), the French Development Agency (AFD), German Agency for International Cooperation (GIZ GmbH), German Federal Ministry for Economic Cooperation and Development (BMZ), Japan International Cooperation Agency (JICA), U.K. Department for International Development (DFID), and U.S. Agency for Inter­ national Development (USAID). The team also held consultations with the advisory board of KCP. The team is grateful to all those who took part in these events. Civil society organizations (CSOs) represented at consultations included, among others, ActionAid, Bill & Melinda Gates Foundation, Education International, Global Campaign for Education, LEGO Foundation, MasterCard Foundation, ONE Campaign, Oxfam, Save the Children, Teach for All, and World Vision. In addition, a diverse group of CSOs participated in a CSO Forum session held during the 2017 World Bank/IMF Spring Meetings and in an e-forum held in March 2017. The team is grateful to these CSOs for their input and useful engagement. Researchers and academics provided helpful feedback at WDR-oriented sessions at the 2016 Research on Improving Systems of Education (RISE) Conference at Oxford Univer- sity, 2017 meetings of the Allied Social Sciences Associations (ASSA), 2017 meetings of the Society for Research on Education Effectiveness (SREE), 2017 Mexico Conference on Politi- cal Economy of Education, and 2017 meeting of the Systems Approach for Better Education Results (SABER) Advisory Panel. In addition, events dedicated to the WDR were organized by the Aga Khan Foundation and Global Affairs Canada in Ottawa; Brookings Center for Universal Education in Washington, DC; Columbia School of International and Public Affairs and Cornell University in New York; Development Policy Forum of GIZ GmbH, on behalf of BMZ, in Berlin; JICA in Tokyo; Université Félix Houphouët-Boigny in Abidjan; and USAID in Washington, DC. This Report draws on background papers prepared by Violeta Arancibia, Felipe Barrera- Osorio, Tessa Bold, Pierre de Galbert, Louise Fox, Dileni Gunewardena, James Habyarimana, Michael Handel, Anuradha Joshi, Kanishka Kacker, Michelle Kaffenberger, Upaasna Kaul, Elizabeth M. King, Gayle Martin, Eema Masood, Ezequiel Molina, Sebastian Monroy- Taborda, Kate Moriarty, Anna Popova, Lant Pritchett, Christophe Rockmore, Andrew Rosser, Maria Laura Sanchez Puerta, Priyam Saraf, M. Najeeb Shafiq, Brian Stacy, Jakob Svensson, Namrata Tognatta, Robert Toutkoushian, Michael Trucano, Waly Wane, Tim Williams, and Attiya Zaidi. viii | Acknowledgments The team drew on the analysis, research, and literature reviews of researchers and specialists from across the world. In addition, the team would like to thank the following for their feedback and suggestions: Christine Adick, Ben Ansell, Manos Antoninis, Caridad Araujo, David Archer, Belinda Archibong, Monazza Aslam, Girindre Beeharry, Penelope Bender, Peter Bergman, Raquel Bernal, Robert Birch, Tarsald Brautaset, Barbara Bruns, Annika Calov, Michael Clemens, Luis Crouch, Rohen d’Aiglepierre, Rossieli Soares da Silva, Momar Dieng, Rob Doble, Amy Jo Dowd, Margaret Dubeck, Sandra Dworack, Alex Eble, Marcel Fafchamps, John Floreta, Eli Friedman, Akihiro Fushimi, Paul Gertler, Rachel Glennerster, Paul Glewwe, Amber Gove, Oliver Haas, James Habyarimana, Jeffrey Hammer, Michael Handel, Christoph Hansert, Blanca Heredia, Sam Hickey, Veronika Hilber, Arja-Sisko Holappa, Naomi Hossain, Huang Xiaoting, Ali Inam, Dhir Jhingran, Emmanuel Jimenez, Maciej Jubowski, Ravi Kanbur, Cheikh Kane, Jouni Kangasniemi, Devesh Kapur, Vishnu Karki, Nina Kataja, Venita Kaul, Kim Kerr, Elizabeth M. King, Kenneth King, Geeta Kingdon, Eiji Kozuka, Michael Kremer, K. P. Krishnan, Kazuo Kuroda, Elina Lehtomäki, Henry Levin, Brian Levy, Krystelle Lochard, Karen Macours, Lu Mai, Akshay Mangla, M. A. Mannan, Santhosh Mathew, Imran Matin, Jordan Matsudaira, Karthik Muralidharan, Essa Chanie Mussa, Charles Nelson III, Aromie Noe, Munaz Ahmed Noor, Mario Novelli, Mead Over, Jan Pakulski, Benjamin Piper, Lant Pritchett, Ritva Reinikka, Risto Rinne, Jo Ritzen, Francisco Rivera Batiz, John Rogers, Caine Rolleston, Andrew Rosser, David Sahn, Justin Sandefur, Yasuyuki Sawada, Andreas Schleicher, Ben Ross Schneider, Dorothea Schonfeld, Olaf Seim, Abhijeet Singh, David Skinner, William Smith, Prachi Srivastava, Liesbet Steer, R. Subrahmanyam, Sudarno Sumarto, Jan Svejnar, Jakob Svensson, Soubhy Tawil, Valerie Tessio, Auli Toom, Miguel Urqiola, Jouni Välijärvi, Olli Vesterinen, Joseph Wales, Libing Wang, Michael Ward, Kevin Watkins, Mark Wenz, Yang Po, Khair Mohamad Yusof, and Andrew Zeitlin. Team members also drew heavily on their own experiences and interactions with the many dedicated educators, administrators, and policy makers who work in often difficult conditions to provide students with the best educational opportunities possible. A number of World Bank colleagues provided insightful comments, feedback, and collaboration: Junaid Ahmad, Omar Arias, Nina Arnhold, Ana Belver, Hana Brixi, James Brumby, Pedro Cerdan Infantes, Marie-Hélène Cloutier, Aline Coudouel, Amit Dar, Jishnu Das, Amanda Epstein Devercelli, Gregory Elacqua, Emanuela Galasso, Diana Hincapie, Alaka Holla, Peter Holland, Sachiko Kataoka, Stuti Khemani, Igor Kheyfets, Kenneth King, Eva Kloeve, Steve Knack, Xiaoyan Liang, Toby Linden, Oni Lusk-Stover, Francisco Mar­ molejo, Yasuhiko Matsuda, Julie McLaughlin, Muna Meky, Ezequiel Molina, Caitlin Moss, Matiullah Noori, Anna Olefir, Owen Ozier, Andrew Ragatz, Vijayendra Rao, Dan Rogger, Audrey Sacks, María Laura Sánchez Puerta, Indhira Santos, William Seitz, Shabnam Sinha, Lars Sondergaard, Dewi Susanti, Christopher Thomas, Michael Trucano, Adam Wagstaff, and Melanie Walker. The team would also like to thank the World Bank colleagues who helped organize and facilitate consultations and advised on translations: Gabriela Geraldes Bastos, Paolo Belli, Moussa Blimpo, Andreas Blom, Leandro Costa, Oumou Coulibaly, Meaza Zerihun Demissie, Safaa El-Kogali, Tazeen Fasih, Ning Fu, Elena Glinskaya, Marek Hanusch, Pimon Iamsri- pong, Susiana Iskandar, Nalin Jena, Hamoud Abdel Wedoud Kamil, Adriane Landwehr, Dilaka Lathapipat, Khady Fall Lo, Norman Loayza, André Loureiro, Hope Nanshemeza, Mademba Ndiaye, Koichi Omori, Azedine Ouerghi, Tigran Shmis, Taleb Ould Sid’ahmed, Lars Sondergaard, Dewi Susanti, Yasusuke Tsukagoshi, and Michael Woolcock. In addition, the team is grateful to the many World Bank colleagues who provided writ- ten comments during the formal Bankwide review process: Cristian Aedo, Inga Afanasieva, Ahmad Ahsan, Edouard Al Dahdah, Umbreen Arif, Nina Arnhold, Anna Autio, Arup Banerji, Elena Bardasi, Sajitha Bashir, Ana Belver, Raja Bentaouet Kattan, Luis Benveniste, Moussa Blimpo, Erik Bloom, Vica Bogaerts, Susan Caceres, César Calderón, Ted Haoquan Chu, Punam Chuhan-Pole, Fernando Ramirez Cortes, Michael Crawford, Laisa Daza, Bénédicte de la Brière, Gabriel Demombynes, Shanta Devarajan, Sangeeta Dey, Ousmane Diagana, Acknowledgments | ix Ousmane Dione, Safaa El Tayeb El-Kogali, Marianne Fay, María Marta Ferreyra, Carina Fonseca, Marie Gaarder, Roberta Gatti, Ejaz Syed Ghani, Elena Glinskaya, Markus Goldstein, Melinda Good, David Gould, Sangeeta Goyal, Caren Grown, Keith Hansen, Amer Hasan, Caroline Heider, Katia Herrera, Niels Holm-Nielsen, Dingyong Hou, Elena Ianchovichina, Keiko Inoue, Sandeep Jain, Omer Karasapan, Michel Kerf, Asmeen Khan, Igor Kheyfets, Youssouf Kiendrebeogo, Daniel John Kirkwood, Eva Kloeve, Markus Kostner, Daniel Lederman, Hans Lofgren, Gladys López-Acevedo, Javier Luque, Michael Mahrt, Francisco Mar­ molejo, Kris McDonall, Mahmoud Mohieldin, Lili Mottaghi, Mary Mulusa, Yoko Nakashima, Shiro Nakata, Muthoni Ngatia, Shinsaku Nomura, Dorota Agata Nowak, Michael O’Sullivan, Arunma Oteh, Aris Panou, Georgi Panterov, Suhas Parandekar, Harry Patrinos, Dhushyanth Raju, Martín Rama, Sheila Redzepi, Lea Marie Rouanet, Jaime Saavedra, Hafida Sahraoui, Sajjad Shah, Sudhir Shetty, Mari Shojo, Lars Sondergaard, Nikola Spatafora, Venkatesh Sundararaman, Janssen Teixeira, Jeff Thindwe, Hans Timmer, Yvonne Tsikata, Laura Tuck, Anuja Utz, Julia Valliant, Axel van Trotsenburg, Carlos Vegh, Binh Thanh Vu, Jan Walliser, Jason Weaver, Michel Welmond, Deborah Wetzel, Christina Wood, and Hanspeter Wyss. The team apologizes to any individuals or organizations inadvertently omitted from this list and expresses its gratitude to all who contributed to this Report, including those whose names may not appear here. The team members would also like to thank their families for their support throughout the preparation of this Report. And finally, the team members thank the many children and youth who have inspired them through interactions in classrooms around the world over the years—as well as the many others whose great potential has motivated this Report. The World Development Report 2018 is dedicated to them. x | Acknowledgments OVERVIEW  earning to realize L education’s promise Learning to realize education’s promise Assess learning Act on evidence Align actors to make it a serious goal to make schools to make the whole work for all learners system work for learning OVERVIEW Learning to realize education’s promise “Education is the most powerful weapon we can use to change the world.” NELSON MANDELA (2003) If your plan is for one year, plant rice. If your plan is for ten years, plant trees. “ If your plan is for one hundred years, educate children.” KUAN CHUNG (7TH CENTURY BC) Schooling is not the same as learning. In Kenya, transaction, reading a doctor’s instructions, or inter- Tanzania, and Uganda, when grade 3 students were preting a campaign promise—let alone building a asked recently to read a sentence such as “The fulfilling career or educating their children. name of the dog is Puppy,” three-quarters did not This learning crisis is a moral crisis. When deliv- understand what it said.1 In rural India, just under ered well, education cures a host of societal ills. three-quarters of students in grade 3 could not solve For individuals, it promotes employment, earnings, a two-digit subtraction such as 46 – 17, and by grade health, and poverty reduction. For societies, it spurs 5 half could still not do so.2 Although the skills of innovation, strengthens institutions, and fosters Brazilian 15-year-olds have improved, at their cur- social cohesion. But these benefits depend largely rent rate of improvement they won’t reach the rich- on learning. Schooling without learning is a wasted country average score in math for 75 years. In reading, opportunity. More than that, it is a great injustice: it will take more than 260 years.3 Within countries, the children whom society is failing most are the ones learning outcomes are almost always much worse who most need a good education to succeed in life. for the disadvantaged. In Uruguay, poor children in Any country can do better if it acts as though learn- grade 6 are assessed as “not competent” in math at ing really matters. That may sound obvious—after five times the rate of wealthy children.4 Moreover, all, what else is education for? Yet even as learning such data are for children and youth lucky enough to goals are receiving greater rhetorical support, in be in school. Some 260 million aren’t even enrolled in practice many features of education systems conspire primary or secondary school.5 against learning. This Report argues that countries These countries are not unique in the challenges can improve by advancing on three fronts: they face. (In fact, they deserve credit for measuring student learning and making the results public.) Assess learning—to make it a serious goal. This •  Worldwide, hundreds of millions of children reach means using well-designed student assessments young adulthood without even the most basic life to gauge the health of education systems (not skills. Even if they attend school, many leave without primarily as tools for administering rewards and the skills for calculating the correct change from a punishments). It also means using the resulting Overview | 3 learning measures to spotlight hidden exclu- The three dimensions of the sions, make choices, and evaluate progress. • Act on evidence—to make schools work for all learners. learning crisis Evidence on how people learn has exploded in Education should equip students with the skills they recent decades, along with an increase in edu- need to lead healthy, productive, meaningful lives. cational innovation. Countries can make much Different countries define skills differently, but better use of this evidence to set priorities for all share some core aspirations, embodied in their their own practice and innovations. curriculums. Students everywhere must learn how • Align actors—to make the whole system work for to interpret many types of written passages—from learning. Countries must recognize that all the medication labels to job offers, from bank statements classroom innovation in the world is unlikely to to great literature. They have to understand how have much impact if, because of technical and numbers work so that they can buy and sell in mar- political barriers, the system as a whole does not kets, set family budgets, interpret loan agreements, support learning. By taking into account these or write engineering software. They require the real-world barriers and mobilizing everyone who higher-order reasoning and creativity that builds on has a stake in learning, countries can support these foundational skills. And they need the socio- innovative educators on the front lines. emotional skills—such as perseverance and the ability to work on teams—that help them acquire and apply When improving learning becomes a priority, the foundational and other skills. great progress is possible. In the early 1950s, the Many countries are not yet achieving these goals. Republic of Korea was a war-torn society held back by First, the learning that one would expect to happen very low literacy levels. By 1995 it had achieved uni- in schools—whether expectations are based on formal versal enrollment in high-quality education through curriculums, the needs of employers, or just common secondary school. Today, its young people perform at sense—is often not occurring. Of even greater con- the highest levels on international learning assess- cern, many countries are failing to provide learning ments. Vietnam surprised the world when the 2012 for all. Individuals already disadvantaged in society— results of the Programme for International Student whether because of poverty, location, ethnicity, gen- Assessment (PISA) showed that its 15-year-olds were der, or disability—learn the least. Thus education performing at the same level as those in Germany— systems can widen social gaps instead of narrowing even though Vietnam was a lower-middle-income them. What drives the learning shortfalls is becoming country. Between 2009 and 2015, Peru achieved some clearer thanks to new analyses spotlighting both the of the fastest growth in overall learning outcomes—an immediate cause—poor service delivery that ampli- improvement attributable to concerted policy action. fies the effects of poverty—and the deeper system- In Liberia, Papua New Guinea, and Tonga, early grade level problems, both technical and political, that allow reading improved substantially within a very short poor-quality schooling to persist. time thanks to focused efforts based on evidence. And recently, Malaysia and Tanzania launched promising Learning outcomes are poor: Low levels, societywide collaborative approaches to systemati- high inequality, slow progress cally improving learning. The recent expansion in education Progress like this requires a clear-eyed diagnosis, is impressive by historical stan- followed by concerted action. Before showing what dards. In many developing coun- can be done to fulfill education’s promise, this over- tries over the last few decades, net view first shines a light on the learning crisis: how enrollment in education has greatly and why many countries are not yet achieving “learn- outpaced the historic performance ing for all.” This may make for disheartening reading, of today’s industrial countries. For Problem but it should not be interpreted as saying that all is example, it took the United States dimension 1: lost—only that too many young people are not getting 40 years—from 1870 to 1910—to Outcomes the education they need. The rest of the overview increase girls’ enrollments from 57 shows how change is possible if systems commit to percent to 88 percent. By contrast, Morocco achieved “all for learning,” drawing on examples of families, a similar increase in just 11 years.6 The number of educators, communities, and systems that have made years of schooling completed by the average adult in real progress. the developing world more than tripled from 1950 to 4 | World Development Report 2018 Figure O.1 Shortfalls in learning start early Percentage of grade 2 students who could not perform simple reading or math tasks, selected countries a. Grade 2 students who could not read b. Grade 2 students who could not a single word of a short text perform two-digit subtraction 100 100 80 80 60 60 Percent Percent 40 40 20 20 0 0 . a q co ia ia an ep a na da a da na ua q a i i al aw aw bi Ira di di Ira ny er an oc rd ,R ep ag ha ha an an m In In al Lib Ke al nz Jo or Za N G G en ar Ug Ug M M M Ta ic m N Ye Sources: WDR 2018 team, using reading and mathematics data for Kenya and Uganda from Uwezo, Annual Assessment Reports, 2015 (http://www.uwezo .net/); reading and mathematics data for rural India from ASER Centre (2017); reading data for all other countries from U.S. Agency for International Development (USAID), Early Grade Reading Barometer, 2017, accessed May 30, 2017 (http://www.earlygradereadingbarometer.org/); and mathematics data for all other countries from USAID/RTI Early Grade Mathematics Assessment intervention reports, 2012–15 (https://shared.rti.org/sub-topic/early -grade-math-assessment-egma). Data at http://bit.do/WDR2018-Fig_O-1. Note: These data typically pertain to selected regions in the countries and are not necessarily nationally representative. Data for India pertain to rural areas. 2010, from 2.0 to 7.2 years.7 By 2010 the average worker of grade 6 students in Southern and East Africa were in Bangladesh had completed more years of school- able to go beyond the level of simply deciphering ing than the typical worker in France in 1975.8 This words, and less than 40 percent got beyond basic progress means that most enrollment gaps in basic numeracy.14 Among grade 6 students in West and education are closing between high- and low-income Central Africa in 2014, less than 45 percent reached countries. By 2008 the average low-income country the “sufficient” competency level for continuing stud- was enrolling students in primary school at nearly ies in reading or mathematics—for example, the rest the same rate as the average high-income country. could not answer a math problem that required them But schooling is not the same as learning.9 Chil- to divide 130 by 26.15 In rural India in 2016, only half of dren learn very little in many education systems grade 5 students could fluently read text at the level around the world: even after several years in school, of the grade 2 curriculum, which included sentences millions of students lack basic literacy and numeracy (in the local language) such as “It was the month skills. In recent assessments in Ghana and Malawi, of rains” and “There were black clouds in the sky.”16 more than four-fifths of students at the end of grade 2 These severe shortfalls constitute a learning crisis. were unable to read a single familiar word such Although not all developing countries suffer from as the or cat (figure O.1).10 Even in Peru, a middle- such extreme shortfalls, many are far short of the lev- income country, that share was half before the recent els they aspire to. According to leading international reforms.11 When grade 3 students in Nicaragua were assessments of literacy and numeracy—Progress in tested in 2011, only half could correctly solve 5 + 6.12 International Reading Literacy Study (PIRLS) and In urban Pakistan in 2015, only three-fifths of grade 3 Trends in International Mathematics and Science students could correctly perform a subtraction such Study (TIMSS)—the average student in low-income as 54 – 25, and in rural areas only just over two-fifths countries performs worse than 95 percent of the stu- could.13 dents in high-income countries, meaning that student This slow start to learning means that even stu- would be singled out for remedial attention in a class dents who make it to the end of primary school do not in high-income countries.17 Many high-performing master basic competencies. In 2007, the most recent students in middle-income countries—young men year for which data are available, less than 50 percent and women who have risen to the top quarter of Overview | 5 Figure O.2 In several countries, the 75th percentile of PISA test takers performs below the 25th percentile of the OECD average Performance of 25th, 50th, and 75th percentiles in 2015 PISA mathematics assessment, selected countries Percentile 75th 600 50th 550 500 25th Mathematics score 450 400 350 300 ic Ko a Tu o sia YR do il sia an ol ru ta a av m Fe age n nd y a, d p. ng an e ca a z an or tio v ri i n tvi bl a C mb Pe Re a so Ri rd la Ko nla p ,F ge ni ne EC ietn ap pu La Br ra m er Ja Ire Jo o ia Al er Fi de Re V os on re G In C Si D an ed n ic ac ia O in ss M m Ru Do OECD interquartile range Source: WDR 2018 team, using data from Programme for International Student Assessment (PISA) 2015 (OECD 2016). Data at http://bit.do/WDR2018-Fig_O-2. their cohorts—would rank in the bottom quarter in a income countries as well, with disadvantaged stu- wealthier country. In Algeria, the Dominican Republic, dents greatly overrepresented among the low scorers. and Kosovo, the test scores of students at the cutoff Costa Rica and Qatar have the same average score on for the top quarter of students (the 75th percentile of one internationally benchmarked assessment (TIMSS the distribution of PISA test takers) are well below the 2015)—but the gap between the top and bottom quar- cutoff for the bottom quarter of students (25th per- ters of students is 138 points in Qatar, compared with centile) of Organisation for Economic Co-operation 92 points in Costa Rica. The gap between the top and and Development (OECD) countries (figure O.2). Even bottom quarters in the United States is larger than in Costa Rica, a relatively strong performer in educa- the gap in the median scores between Algeria and the tion, performance at the cutoff for the top quarter of United States. students is equal to performance at the cutoff for the Students often learn little from year to year, but bottom quarter in Germany. early learning deficits are magnified over time. Stu- The learning crisis amplifies inequality: it severely dents who stay in school should be rewarded with hobbles the disadvantaged youth who most need the steady progress in learning, whatever disadvantages boost that a good education can offer. For students in they have in the beginning. And yet in Andhra Pradesh, many African countries, the differences by income India, in 2010, low-performing students in grade 5 level are stark (figure O.3). In a recent assessment were no more likely to answer a grade 1 question cor- (Programme d’Analyse des Systèmes Éducatifs de la rectly than those in grade 2. Even the average student Confemen, PASEC, 2014) administered at the end of in grade 5 had about a 50 percent chance of answering the primary cycle, only 5 percent of girls in Camer- a grade 1 question correctly—compared with about 40 oon from the poorest quintile of households had percent in grade 2.19 In South Africa in the late 2000s, learned enough to continue school, compared with 76 the vast majority of students in grade 4 had mastered percent of girls from the richest quintile.18 Learning only the mathematics curriculum from grade 1; most gaps in several other countries—Benin, the Republic of those in grade 9 had mastered only the mathemat- of Congo, and Senegal—were nearly as wide. Large ics items from grade 5.20 In New Delhi, India, in 2015, gaps among learners afflict many high- and middle- the average grade 6 student performed at a grade 3 6 | World Development Report 2018 Figure O.3 Children from poor households in Africa typically learn much less Percentage of grade 6 PASEC test takers in 2014 who scored above (blue) and below (orange) the sufficiency level on reading achievement: poorest and richest quintiles by gender, selected countries 100 75 50 25 Percent 0 25 50 75 100 F M F M F M F M F M F M F M F M F M F M F M F M F M F M F M F M F M F M Poor Rich Poor Rich Poor Rich Poor Rich Poor Rich Poor Rich Poor Rich Poor Rich Poor Rich Niger Togo Cameroon Congo, Rep. Benin Côte d’Ivoire Burkina Faso Senegal Burundi Not competent Low competency High competency Source: WDR 2018 team, using data from World Bank (2016b). Data at http://bit.do/WDR2018-Fig_O-3. Note: Socioeconomic quintiles are defined nationally. “Not competent” refers to levels 0–2 in the original coding and is considered below the sufficiency level for school continuation; “low competency” refers to level 3; and “high competency” refers to level 4. F = female; M = male; PASEC = Programme d’Analyse des Systèmes Éducatifs de la Confemen. level in math. Even by grade 9, the average student Figure O.4 Students often learn little from year had reached less than a grade 5 level, and the gap to year, and early learning deficits are magnified between the better and worse performers grew over over time time (figure O.4). In Peru and Vietnam—one of the low- Assessed grade-level performance of students relative to enrolled grade, est and one of the highest performers, respectively, on New Delhi, India (2015) the PISA assessment of 15-year-old students—5-year- a. Mathematics b. Language olds start out with similar math skills, but students 9 9 in Vietnam learn much more for each year of school- ing at the primary and lower secondary levels.21 Although some countries are making progress 8 8 on learning, their progress is typically slow. Even the Grade-level performance Grade-level performance middle-income countries that are catching up to the 7 7 top performers are doing so very slowly. Indonesia has registered significant gains on PISA over the last 6 6 10–15 years. And yet, even assuming it can sustain its 2003–15 rate of improvement, Indonesia won’t reach the OECD average score in mathematics for another 5 5 48 years; in reading, for 73. For other countries, the wait could be even longer: based on current trends, 4 4 it would take Tunisia over 180 years to reach the OECD average for math and Brazil over 260 years 3 3 to reach the OECD average for reading. Moreover, 6 7 8 9 6 7 8 9 these calculations are for countries where learning Enrolled grade Enrolled grade has improved. Across all countries participating in Expected performance 75th percentile multiple rounds of PISA since 2003, the median gain Average assessed performance 25th percentile in the national average score from one round to the Source: WDR 2018 team, using data from Muralidharan, Singh, and Ganimian (2016). Data at http://bit.do next was zero. /WDR2018-Fig_O-4. Overview | 7 Figure O.5 The percentage of primary school students who pass a minimum proficiency threshold is often low Median percentage of students in late primary school who score above a minimum proficiency level on a learning assessment, by income group and region 100 80 60 Percent 40 20 0 Low- Lower-middle- Upper-middle- High- Sub-Saharan Middle East Latin East Asia Europe and income income income income Africa and America and Pacific Central Asia countries countries countries countries North Africa and the Caribbean Mathematics Reading Source: WDR 2018 team, using “A Global Data Set on Education Quality” (2017), made available to the team by Nadir Altinok, Noam Angrist, and Harry Anthony Patrinos. Data at http://bit.do /WDR2018-Fig_O-5. Note: Bars show the unweighted cross-country median within country grouping. Regional averages exclude high-income countries. India and China are among the countries excluded for lack of data. Minimum proficiency in mathematics is benchmarked to the Trends in International Mathematics and Science Study (TIMSS) assessment and in reading to the Progress in International Reading Literacy Study (PIRLS) assessment. Minimum proficiency in mathematics means that students have some basic mathematical knowledge such as adding or subtracting whole numbers, recognizing familiar geometric shapes, and reading simple graphs and tables (Mullis and others 2016). Minimum proficiency in reading means that students can locate and retrieve explicitly stated detail when reading literary texts and can locate and reproduce explicitly stated information from the beginning of informational texts (Mullis and others 2012). Because of this slow progress, more than 60 per- school. In 2016, 61 million children of primary school cent of primary school children in developing coun- age—10 percent of all children in low- and lower- tries still fail to achieve minimum proficiency in learn- middle-income countries—were not in school, along ing, according to one benchmark. No single learning with 202 million children of secondary school age.24 assessment has been administered in all countries, Children in fragile and conflict-affected countries but combining data from learning assessments in 95 accounted for just over a third of these, a dispropor- countries makes it possible to establish a globally com- tionate share. In the Syrian Arab Republic, which parable “minimum proficiency” threshold in math.22 achieved universal primary enrollment in 2000, the Below this threshold, students have not mastered even civil war had driven 1.8 million children out of school basic mathematical skills, whether making simple by 2013.25 Almost all developing countries still have computations with whole numbers, using fractions pockets of children from excluded social groups who or measurements, or interpreting simple bar graphs. do not attend school. Poverty most consistently pre- In high-income countries, nearly all students—99 per- dicts failing to complete schooling, but other charac- cent in Japan, 98 percent in Norway, 91 percent in Aus- teristics such as gender, disability, caste, and ethnicity tralia—achieve this level in primary school.23 But in also frequently contribute to school participation other parts of the world the share is much lower: just shortfalls (figure O.6). 7 percent in Mali, 30 percent in Nicaragua, 34 percent But it’s not just poverty and conflict that keep in the Philippines, and 76 percent in Mexico. In low- children out of school; the learning crisis does, too. income countries, 14 percent of students reach this When poor parents perceive education to be of low level near the end of primary school, and in lower- quality, they are less willing to sacrifice to keep their middle-income countries 37 percent do (figure O.5). children in school—a rational response, given the Even in upper-middle-income countries only 61 per- constraints they face.26 Although parental perceptions cent reach this minimum proficiency. of school quality depend on various factors, from the The ultimate barrier to learning is no schooling at physical condition of schools to teacher punctuality, all—yet hundreds of millions of youth remain out of parents consistently cite student learning outcomes 8 | World Development Report 2018 Figure O.6 School completion is higher for richer and urban families, but gender gaps are more context-dependent Gaps in grade 6 completion rates (percent) for 15- to 19-year-olds, by wealth, location, and gender a. Richest–poorest b. Urban–rural c. Male–female 80 80 80 Percentage point gap between Percentage point gap between Percentage point gap between richest and poorest quintiles 60 60 60 male and female urban and rural 40 40 40 20 20 20 0 0 0 −20 −20 −20 20 40 60 80 100 20 40 60 80 100 20 40 60 80 100 Overall grade 6 completion rate (%) Overall grade 6 completion rate (%) Overall grade 6 completion rate (%) Source: WDR 2018 team, using data from Filmer (2016). Data at http://bit.do/WDR2018-Fig_O-6. Note: The data presented are the latest available by country, 2005–14. Each vertical line indicates the size and direction of the gap for a country. as a critical component.27 These outcomes can affect the foundational cognitive skills are essential, and behavior: holding student ability constant, students systems cannot bypass the challenges of developing in the Arab Republic of Egypt who attended poorer- them as they target higher-order skills. performing schools were more likely to drop out.28 Tackling the learning crisis and skills gaps requires Learning shortfalls during the school years even- diagnosing their causes—both their immediate causes tually show up as weak skills in the workforce. Thus at the school level and their deeper systemic drivers. the job skills debate reflects the learning crisis. Work Given all the investments countries have made in skill shortages are often discussed in a way that is education, shortfalls in learning are discouraging. But disconnected from the debate on learning, but the one reason for them is that learning has not always two are parts of the same problem. Because education received the attention it should have. As a result, systems have not prepared workers adequately, many stakeholders lack actionable information about what enter the labor force with inadequate skills. Measur- is going wrong in their schools and in the broader ing adult skills in the workplace is hard, but recent society, and so they cannot craft context-appropriate initiatives have assessed a range of skills in the adult responses to improve learning. Acting effectively populations of numerous countries. They found that requires first understanding how schools are failing even foundational skills such as literacy and numer- learners and how systems are failing schools. acy are often low, let alone the more advanced skills. The problem isn’t just a lack of trained workers; it is Schools are failing learners a lack of readily trainable workers. Accordingly, many Struggling education systems lack workers end up in jobs that require minimal amounts one or more of four key school-level of reading or math.29 Lack of skills reduces job quality, ingredients for learning: prepared earnings, and labor mobility. learners, effective teaching, learning- The skills needed in labor markets are multi­ focused inputs, and the skilled man- dimensional, so systems need to equip students with agement and governance that pulls far more than just reading, writing, and math—but them all together (figure O.7). The students cannot leapfrog these foundational skills. next section looks at why these links Problem break down; here the focus is on how dimension 2: Whether as workers or members of society, peo- Immediate ple also need higher-order cognitive skills such as they break down. causes problem-solving. In addition, they need socioemo- First, children often arrive in tional skills—sometimes called soft or noncognitive school unprepared to learn—if they skills—such as conscientiousness. Finally, they need arrive at all. Malnutrition, illness, low parental technical skills to perform a specific job. That said, investments, and the harsh environments associated Overview | 9 Figure O.7 Why learning doesn’t happen: Second, teachers often lack the skills or motiva- Four immediate factors that break down tion to be effective. Teachers are the most important factor affecting learning in schools. In the United States, students with great teachers advance 1.5 grade levels or more over a single school year, compared te d with just 0.5 grade levels for those with an ineffective iva Un ot pr teacher.34 In developing countries, teacher quality m le ep un s ar can matter even more than in wealthier countries.35 er n ch d ar s But most education systems do not attract applicants an ed er a l ed with strong backgrounds. For example, 15-year-old te kil students who aspire to be teachers score below the Uns national average on PISA in nearly all countries.36 Beyond that, weak teacher education results in teach- ers lacking subject knowledge and pedagogical skills. LEARNING In 14 Sub-Saharan countries, the average grade 6 teacher performs no better on reading tests than do the highest-performing students from that grade.37 In S c h at Indonesia, 60 percent of the time in a typical mathe- ts o o do a t h hin matics class is spent on lecturing, with limited time te a pu ng ar ct l an e m remaining for practical work or problem-solving.38 ni in c ag l o o n’ a sn Meanwhile, in many developing countries substantial e le em t ’t h g nd a ent S c t do an d le e a amounts of learning time are lost because classroom ar ct th ing nin ch time is spent on other activities or because teachers g te a are absent. Only a third of total instructional time was used in Ethiopia, Ghana, and Guatemala.39 Across seven African countries, one in five teachers was Source: WDR 2018 team. absent from school on the day of an unannounced visit by survey teams, with another fifth absent from the classroom even though they were at school with poverty undermine early childhood learning.30 (figure O.9).40 The problems are even more severe in Severe deprivations—whether in terms of nutri- remote communities, amplifying the disadvantages tion, unhealthy environments, or lack of nurture by already facing rural students. Such diagnostics are caregivers—have long-lasting effects because they not intended to blame teachers. Rather, they call impair infants’ brain development.31 Thirty percent attention to how systems undermine learning by of children under 5 in developing countries are failing to support them. physically stunted, meaning they have low height Third, inputs often fail to reach classrooms or to for their age, typically due to chronic malnutrition.32 affect learning when they do. Public discourse often The poor developmental foundations and lower levels equates problems of education quality with input of preschool skills resulting from deprivation mean gaps. Devoting enough resources to education is cru- many children arrive at school unprepared to benefit cial, and in some countries resources have not kept fully from it (figure O.8).33 So even in a good school, pace with the rapid jumps in enrollment. For several deprived children learn less. Moreover, breaking reasons, however, input shortages explain only a small out of lower learning trajectories becomes harder as part of the learning crisis. First, looking across systems these children age because the brain becomes less and schools, similar levels of resources are often asso- malleable. Thus education systems tend to amplify ciated with vast differences in learning outcomes.41 initial differences. Moreover, many disadvantaged Second, increasing inputs in a given setting often has youth are not in school. Fees and opportunity costs small effects on learning outcomes.42 Part of the rea- are still major financial barriers to schooling, and son is that inputs often fail to make it to the front lines. social dimensions of exclusion—for example, those A decade ago in Sierra Leone, for example, textbooks associated with gender or disability—exacerbate the were distributed to schools, but follow-up inspec- problem. These inequalities in school participation tions found most of them locked away in cupboards, further widen gaps in learning outcomes. unused.43 Similarly, many technological interventions 10 | World Development Report 2018 Figure O.8 Socioeconomic gaps in cognitive achievement grow with age—even in preschool years Percentage of children ages 3–5 who can recognize 10 letters of the alphabet, by wealth quintile, selected countries a. Central African Republic b. Kazakhstan c. Tunisia 100 100 100 80 80 80 60 60 60 Percent Percent Percent 40 40 40 20 20 20 0 0 0 3 3.5 4 4.5 5 3 3.5 4 4.5 5 3 3.5 4 4.5 5 Age (years) Age (years) Age (years) Richest quintile Poorest quintile 95% confidence interval 95% confidence interval Source: WDR 2018 team, using data from Multiple Indicator Cluster Surveys (http://mics.unicef.org/). Data are for 2010 for the Central African Republic, 2010–11 for Kazakhstan, and 2012 for Tunisia. Data at http://bit.do/WDR2018-Fig_O-8. fail before they reach classrooms, and even when they Figure O.9 In Africa, teachers are often do make it to classrooms, they often do not enhance absent from school or from classrooms teaching or learning. In Brazil, a One Laptop Per Child while at school initiative in several states faced years of delays. Then, Percentage of teachers absent from school and from class on even a year after the laptops finally made it to class- the day of an unannounced visit, participating countries rooms, more than 40 percent of teachers reported never or rarely using them in classroom activities.44 60 Fourth, poor management and governance often undermine schooling quality. Although effective 40 school leadership does not raise student learning Percent directly, it does so indirectly by improving teaching quality and ensuring effective use of resources.45 20 Across eight countries that have been studied, a 1.00 standard deviation increase in an index of manage- ment capacity—based on the adoption of 20 man- 0 agement practices—is associated with a 0.23–0.43 13 14 0 12 4 3 11 3 01 01 01 01 20 20 20 20 standard deviation increase in student outcomes. ,2 ,2 ,2 ,2 a, e, a, l, ia ia go ia ga qu nd ny an an er But school management capacity tends to be lowest To ne Ke bi a ig nz nz Ug Se am N Ta Ta in those countries with the lowest income levels, oz M classroom and management capacity is substantially lower in Teachers absent from the school schools than in manufacturing (figure O.10).46 Inef- fective school leadership means school principals Source: Bold and others (2017). Data at http://bit.do/WDR2018-Fig_O-9. are not actively involved in helping teachers solve Note: “Absent from the classroom” combines absences from school with absences from class among teachers who are at school. Data are from the problems, do not provide instructional advice, and World Bank’s Service Delivery Indicators (SDI) surveys (http://www do not set goals that prioritize learning. School gover- .worldbank.org/sdi). nance—particularly the decision-making autonomy of schools, along with the oversight provided by autonomy, and community engagement fails to affect parents and communities—serves as the framework what happens in classrooms.47 for seeking local solutions and being accountable for Because these quality problems are concentrated them. In many settings, schools lack any meaningful among disadvantaged children, they amplify social Overview | 11 Figure O.10 Management capacity typically disadvantage marginalized communities, is low in schools in low- and middle- but also that resources are used less effectively there, income countries exacerbating the problem. Public policy thus has the effect of widening social gaps rather than offering all Distribution of management scores by sector, participating countries children an opportunity to learn. Systems are failing schools Tanzania Viewed from a systems perspec- tive, the low level of learning and India skills should come as no surprise. Technical complexities and polit- ical forces constantly pull edu- Haiti cation systems out of alignment with learning (figure O.11). Problem Italy dimension 3: Deeper causes Technical challenges: Brazil Reorienting toward learning is hard Complex systems and limited management capacity are obstacles to orienting all parts of an education Mexico system toward learning. First, the various parts of the system need to be aligned toward learning. But Germany actors in the system have other goals—some stated, some not. Promoting learning is only one of these, and not necessarily the most important one. At times, Canada these other goals can be harmful, such as when con- struction firms and bureaucrats collude to provide substandard school buildings for their financial gain. United States At other times, these goals may be laudable, such as nurturing shared national values. But if system ele- Sweden ments are aligned toward these other goals, they will sometimes be at cross-purposes with learning. Even when countries want to prioritize learning, United Kingdom they often lack the metrics to do so. Every system assesses student learning in some way, but many 1 2 3 4 5 systems lack the reliable, timely assessments needed Management score to provide feedback on innovations. For example, is Education Manufacturing a new teacher training program actually making teachers more effective? If the system lacks reliable Sources: Bloom and others (2014, 2015); Lemos and Scur (2016), with updates. information on the quality of teaching and the learn- Data at http://bit.do/WDR2018-Fig_O-10. ing of primary students—comparable across time or Note: The underlying distributions for the education data are shown as classrooms—there is no way to answer that question. bars; for both sectors, the smoothed distributions are shown as curves. The indexes are constructed from the nine items that are comparable across To be truly aligned, parts of the education system sectors. Data on manufacturing are not available for Haiti. also have to be coherent with one another. Imagine that a country has set student learning as a top prior- inequalities. In low-income countries, on average, ity and that it has in place reasonable learning metrics. stunting rates among children under 5 are almost It still needs to leap a major technical hurdle, how- three times higher in the poorest quintile than in ever: ensuring that system elements work together. the richest.48 In schools, problems with teacher If a country adopts a new curriculum that increases absenteeism, lack of inputs, and weak management emphasis on active learning and creative thinking, are typically severest in communities that serve the that alone will not change much. Teachers need to poorest students. It’s not just that spending patterns be trained so that they can use more active learning 12 | World Development Report 2018 methods, and they need to care enough to make the Figure O.11 Technical and political factors divert change because teaching the new curriculum may schools, teachers, and families from a focus on be much more demanding than the old rote learning learning methods. Even if teachers are on board with curricu- lum reform, students could weaken its effects if an unreformed examination system creates misaligned incentives. In Korea, the high-stakes exam system for university entrance has weakened efforts to rs Le e ar reorient secondary school learning. The curriculum ch ne a has changed to build students’ creativity and socio- Te rs emotional skills, but many parents still send their chil- dren to private “cram schools” for test preparation.49 The need for coherence makes it risky to borrow system elements from other countries. Education pol- icy makers and other experts often scrutinize systems LEARNING that have better learning outcomes to identify what they could borrow. Indeed, in the 2000s the search for the secret behind Finland’s admirable record of learn- Sch ing with equity led to a swarm of visiting delegations in what the Finns dubbed “PISA tourism.” Finland’s oo ts pu system gives considerable autonomy to its well- lm na in educated teachers, who can tailor their teaching to l a ge h oo the needs of their students. But lower-performing me Sc systems that import Finland’s teacher autonomy nt into their own contexts are likely to be disappointed: if teachers are poorly educated, unmotivated, and loosely managed, giving them even more autonomy Source: WDR 2018 team. will likely make matters worse. South Africa discov- ered this in the 1990s and 2000s when it adopted a curriculum approach that set goals but left implemen- politicians and teachers happy than on promoting tation up to teachers.50 The approach failed because student learning, or they may simply try to protect it proved to be a poor fit for the capacity of teachers their own positions. Some private suppliers of edu- and the resources at their disposal.51 Home-grown, cation services—whether textbooks, construction, or context-specific solutions are important. schooling—may, in the pursuit of profit, advocate pol- Successful systems combine both alignment and icy choices not in the interest of students. Teachers coherence. Alignment means that learning is the goal and other education professionals, even when moti- of the various components of the system. Coherence vated by a sense of mission, also may fight to main- means that the components reinforce each other in tain secure employment and to protect their incomes. achieving whatever goals the system has set for them. None of this is to say that education actors don’t care When systems achieve both, they are much more likely about learning. Rather, especially in poorly managed to promote student learning. Too much misalignment systems, competing interests may loom larger than or incoherence leads to failure to achieve learning, the learning-aligned interests (table O.2). though the system might achieve other goals (table O.1). Misalignments aren’t random. Because of these competing interests, the choice of a particular policy Political challenges: Key players don’t is rarely determined by whether it improves learning. always want to prioritize student learning More often, the choice is made by the more powerful Political challenges compound technical ones. Many actors in the policy arena. Agents are accountable to education actors have different interests, again one another for different reasons, not just learning. beyond learning. Politicians act to preserve their Given these interests, it should come as no surprise positions in power, which may lead them to target that little learning often results. particular groups (geographic, ethnic, or economic) One problem is that activities to promote learning for benefits. Bureaucrats may focus more on keeping are difficult to manage. Teaching and learning in the Overview | 13 Table O.1 Alignment and coherence both matter Are system Coherent? elements . . . Yes No Yes High performance: Systems well organized to Incoherent strivers: Systems incoherently oriented promote learning toward learning Examples: High performers at each level Examples: Countries that borrow learning-oriented (Shanghai [China], Finland, Vietnam) “best practice” elements but do not ensure that the Aligned toward various elements are coherent with each other learning? No Coherent nonlearners: Systems well organized to Failed systems: Systems that are not trying to promote a different goal achieve learning or anything else in a coherent Examples: Totalitarian or authoritarian systems way focused on promoting loyalty to the state or nation Examples: Systems in failed states building (Stalin-era USSR, Suharto-era Indonesia); systems that focus on school attainment rather than learning (many systems) Source: WDR 2018 team. Table O.2 Multiple interests govern the actions of education stakeholders Examples of . . . Stakeholders Learning-aligned interests Competing interests Teachers Student learning, professional ethic Employment, job security, salary, private tuitions Principals Student learning, teacher performance Employment, salary, good relations with staff, favoritism Bureaucrats Well-functioning schools Employment, salary, rent-seeking Politicians Well-functioning schools Electoral gains, rent-seeking, patronage Parents and students Student learning, employment of Family employment, family income, graduates outdoing others Judiciary Meaningful right to education Favoritism, rent-seeking Employers Skilled graduates Low taxes, narrowly defined self-interests Nongovernment schools (religious, Innovative, responsive schooling Profit, religious mission, funding nongovernmental, for-profit) Suppliers of educational inputs (e.g., High-quality, relevant inputs Profit, influence textbooks, information technology, buildings) International donors Student learning Domestic strategic interests, taxpayer support, employment Source: WDR 2018 team. classroom involve significant discretion by teachers, at simple, easily collected enrollment data. Similarly, as well as regular and repeated interactions between school construction, cash transfer programs, teacher students and teachers.52 These characteristics, cou- hiring, and school grant programs intended to pled with a dearth of reliable information on learn- expand access are all highly visible, easily monitored ing, make managing learning more difficult than investments. pursuing other goals.53 For example, improvements The potential beneficiaries of better foundational in access to education can be monitored by looking learning—such as students, parents, and employers— 14 | World Development Report 2018 often lack the organization, information, or short- other parts of the system, such as higher education or term incentive to press for change. Parents are usually lifelong learning. In these areas, too, many countries not organized to participate in debates at the system suffer from a lack of attention to outcomes, wide gaps level, and they may lack knowledge of the potential in opportunity, and systemic barriers to resolving gains from different policies to improve learning.54 these problems. They also may worry about the potential ramifica- tions for their children or themselves of opposing Still, there are reasons for hope interests such as teachers, bureaucrats, or politicians. Even in countries that seem stuck in low-learning Students have even less power—except sometimes in traps, some teachers and schools manage to strengthen higher education, where they can threaten demon- learning. These examples may not be sustainable—and strations—and, like parents, they may be unaware of they are not likely to spread systemwide without how little they are learning until they start looking efforts to reorient the system toward learning—but for work. Finally, the business community, even if it systems willing to learn from these outliers can bene- suffers from a shortage of skilled graduates to hire, fit. On a larger scale, some regions within countries are often fails to advocate for quality education, instead more successful in promoting learning, as are some lobbying for lower taxes and spending. By contrast countries at each income level. to these potential beneficiaries of reform, the poten- These examples reveal that higher-level system tial losers tend to be more aware of what is at stake equilibriums exist. But is it possible for a whole for them and, in many cases, better organized to act system to escape the low-learning trap, moving collectively. to a better one? There are at least two reasons for As a result, many systems are stuck in low- optimism. First, as countries innovate to improve learning traps, characterized by low accountabil- learning, they can draw on more systematic knowl- ity and high inequality. These traps bind together edge than ever available before about what can work key stakeholders through informal contracts that at the micro level—the level of learners, classrooms, prioritize other goals such as civil service employ- and schools. A number of interventions, innovations, ment, corporate profits, or reelection, perpetuating and approaches have resulted in substantial gains the low-accountability equilibrium. In better-run in learning. These promising approaches come in systems, actors such as bureaucrats and teachers many flavors—new pedagogical methods, ways to can devote much of their energy to improving ensure that students and teachers are motivated, outcomes for students. But in low-learning traps approaches to school management, technologies to those same actors lack either the incentives or the enhance teaching learning—and they may not pay support needed to focus on learning. Instead, they off in all contexts, but the fact that it is possible to are constantly pressured to deliver other services improve learning outcomes should give hope. These for more powerful players. As actors juggle multiple interventions can provide substantial improvements objectives, relying on each other in an environment in learning: almost one or two grade-equivalents for of uncertainty, low social trust, and risk aversion, it some students.55 Even though successful interven- is often in the interest of each to maintain the status tions cannot be imported wholesale into new con- quo—even if society, and many of these actors, would texts, countries can use them as starting points for be better off if they could shift to a higher-quality their own innovations. equilibrium. Second, some countries have implemented This diagnosis has concentrated on the shortfalls reforms that have led to sustained systemwide in foundational learning, as will the priorities for improvements in learning. Finland’s major education action discussed in the next section. However, this reform in the 1970s famously improved the equity focus should not be interpreted as a statement that of outcomes while also increasing quality, so that by other areas are unimportant. Education systems and the time of the first PISA in 2000, Finland topped the their enabling environment are broader and more assessment. More recently, Chile, Peru, Poland, and complex than this Report can cover, so our priority the United Kingdom have made serious, sustained here is to highlight what can be done most immedi- commitments to reforming the quality of their ately to strengthen the foundations of learning on education systems. In all these countries, learning which all successful systems are built. But both the has improved over time—not always steadily, but diagnosis and the priorities for action are relevant for enough to show that system-level reforms can pay off. Overview | 15 The education systems in Shanghai (China) and Assess learning—to make it a serious goal Vietnam today—and Korea decades ago—show that “What gets measured gets man- it is possible to perform far better than income levels aged.” “Just weighing the pig would predict, thanks to a sustained focus on learning doesn’t make it fatter.” There is with equity. Brazil and Indonesia have made consid- some truth to both of these say- erable progress, despite the challenges of reforming ings. Lack of measurement makes large, decentralized systems. it hard to know where things are, where they are going, and what actions are making any difference. Policy How to realize education’s Knowing these things can provide response 1: Assess promise: Three policy focus and stimulate action. But learning measurement that is too removed responses from action can lead nowhere. The challenge is strik- Learning outcomes won’t change unless education ing a balance—finding the right measures for the systems take learning seriously and use learning as right purposes and implementing them within an a guide and metric. This idea can be summarized as appropriate accountability framework. “all for learning.”56 As this section explains, a com- mitment to all for learning—and thus to learning for Use measurement to shine a light on all—implies three complementary strategies: learning The first step to improving systemwide learning is • Assess learning—to make it a serious goal. Measure and to put in place good metrics for monitoring whether track learning better; use the results to guide action. programs and policies are delivering learning. Cred- • Act on evidence—to make schools work for all learners. ible, reliable information can shape the incentives Use evidence to guide innovation and practice. facing politicians. Most notably, information on stu- • Align actors—to make the whole system work for learn- dent learning and school performance—if presented ing. Tackle the technical and political barriers to in a way that makes it salient and acceptable—fosters learning at scale. healthier political engagement and better service delivery. Information also helps policy makers man- These three strategies depend on one another. age a complex system. Adopting a learning metric without any credible way Measuring learning can improve equity by to achieve learning goals will simply lead to frustra- revealing hidden exclusions. As emphasized at the tion. School-level innovations without a learning outset of this overview, the learning crisis is not just metric could take schools off course, and without the a problem for the society and economy overall; it is system-level support they could prove ephemeral. also a fundamental source of inequities and widening And system-level commitment to learning without gaps in opportunity. But because reliable information school-level innovation, and without learning mea- on learning is so spotty in many education systems, sures to guide the reforms, is unlikely to amount to especially in primary and lower secondary schools, more than aspirational rhetoric. But together, the the way the system is failing disadvantaged children three strategies can create change for the better. is a hidden exclusion.58 Unlike exclusion from school, The potential payoff is huge. When children have a lack of learning is often invisible, making it impos- growth mindset, meaning they understand their own sible for families and communities to exercise their great learning potential, they learn much more than right to quality education. when they believe they are constrained by a fixed These measures of learning will never be the only intelligence.57 Societies have the same opportunity. By guide for educational progress, nor should they be. adopting a social growth mindset—recognizing the Education systems should have ways of tracking barriers to learning, but also the very real opportu- progress toward any goal they set for themselves and nities to break them down—they can make progress their students—not just learning. Systems should also on learning. One overarching priority should be to track the critical factors that drive learning—such as end the hidden exclusion of low learning. This is not learner preparation, teacher skills, quality of school just the right thing to do; it is also the surest way to management, and the level and equity of financing. improve average learning levels and reap education’s But learning metrics are an essential starting point full rewards for society as a whole. for improving lagging systems. 16 | World Development Report 2018 Figure O.12 Many countries lack information on learning outcomes Percentage of countries with data to monitor progress toward the Sustainable Development Goals for learning by the end of primary or lower secondary school 100 80 60 Percent 40 20 0 Primary Lower Primary Lower Primary Lower Primary Lower Primary Lower secondary secondary secondary secondary secondary Latin America Arab states Sub-Saharan Asia and World and the Africa Pacific Caribbean Mathematics Reading Source: UIS (2016). Data at http://bit.do/WDR2018-Fig_O-12. Note: Regional groupings follow UNESCO definitions. There is too little measurement of learning, progress toward the United Nations’ Sustainable not too much Development Goals found that of the 121 countries A recommendation to start tackling the learning cri- studied, a third lack the data required to report on sis with more and better measurement of learning the levels of reading and mathematics proficiency may seem jarring. Many education debates highlight of children at the end of primary school.62 Even the risks of overtesting or an overemphasis on tests. more lack data for the end of lower secondary school In the United States, two decades of high-stakes test- (figure O.12). Even when countries have these data, ing have led to patterns of behavior consistent with they are often from one-off assessments that do not these concerns.59 Some teachers have been found to allow systematic tracking over time. A lack of good concentrate on test-specific skills instead of untested measurement means that education systems are subjects, and some schools have engaged in strategic often flying blind—and without even agreement on behavior to ensure that only the better-performing the destination. students are tested, such as assigning students to special education that excuses them from testing.60 In Use a range of metrics with one ultimate the extreme, problems have expanded to convictions goal for systemic cheating at the school district level.61 At Different learning metrics have different purposes, the same time, media coverage of education in many but each contributes to learning for all. Teachers low- and middle-income countries (and some high- assess students in classrooms every day—formally income ones) often focuses on high-stakes national or informally—even in poorly resourced, poorly man- examinations that screen candidates for tertiary edu- aged school systems. But using metrics properly to cation—raising concerns about an overemphasis on improve learning systemwide requires a spectrum of testing. types of assessment that, together, allow educators But in many systems the problem is too little and policy makers to use the right combination of focus on learning—not too much. Many countries teaching approaches, programs, and policies. lack information on even basic reading and math Formative assessment by teachers helps guide competencies. An assessment of capacity to monitor instruction and tailor teaching to the needs of Overview | 17 students. Well-prepared, motivated teachers do not raise awareness of how a country is falling short of its need to operate in the dark: they know how to assess peers in building human capital. the learning of students regularly, formally and Two other types of learning metrics measured informally. As the next section discusses, this type of in nonschool settings can be used to strengthen the regular check-in is important because many students quality and equity focus of assessment systems. lag so far behind that they effectively stop learning. Grassroots accountability movements—led by civil Knowing where students are allows teachers to society organizations such as the ASER Centre in India adjust their teaching accordingly and to give students and Uwezo in East Africa—have deployed citizen-led learning opportunities they can handle. Singapore assessments that recruit volunteers to measure the has successfully used this approach—identifying foundational learning of young children in their lagging students in grade 1 using screening tests and communities. These organizations then use their then giving them intensive support to bring them up learning data to advocate for education reform. Some to grade level.63 multipurpose household surveys also collect learning National and subnational learning assessments data, enabling researchers to analyze how learning provide system-level insights that classroom assess- outcomes correlate with income and community ments by teachers cannot. To guide an education variables. Both types of assessments are administered system, policy makers need to understand whether in people’s homes, not schools. As a result, they students are mastering the national curriculum, in don’t suffer from a key weakness of school-based which areas students are stronger or weaker, whether assessments: when marginal students drop out, their certain population groups are lagging behind and absence can improve the average scores on school by how much, and which factors are associated with assessments, thereby creating a perverse incentive better student achievement. There is no effective way for school leaders. But household-based assessments to aggregate the results of classroom-level formative yield learning metrics that reward systems for assessment by teachers into this type of reliable improving both access and quality. This is crucial to system-level information. This is why systems need ensuring that no child is written off. Even for students assessments of representative samples of students who are in school, household-based assessments pro- across wider jurisdictions, such as countries or prov- vide an alternative source of learning data, which can inces. Such assessments can be an especially import- be important in settings where official assessments ant part of tracking systemwide progress because are of questionable quality. they are anchored in a system’s own expectations for itself. And national assessments can provide a check Measurement can be hard on the quality of subnational assessments by flagging Why isn’t there more and better measurement of cases in which trends or levels of student achieve- learning? As with system barriers to learning, bar- ment diverge across the two. In the United States, riers to better measurement are both technical and the National Assessment of Educational Progress has political. From a technical perspective, conducting played this role.64 good assessments is not easy. At the classroom International assessments also provide informa- level, teachers lack the training to assess learning tion that helps improve systems. Globally bench- effectively, especially when assessments try to cap- marked student assessments such as PISA, TIMSS, ture higher-order skills—say, through project-based and PIRLS, as well as regionally benchmarked ones assessment—rather than rote learning. And at the such as PASEC in West and Central Africa and the system level, education ministries lack the capacity Latin American Laboratory for Assessment of the to design valid assessments and implement them in a Quality of Education (LLECE), provide an additional sample of schools. Political factors intrude as well. To perspective on how well students are learning. They paraphrase an old saying, policy makers may decide allow assessment of country performance in a way it is better to avoid testing and be assumed inef- that is comparable across countries, and they pro- fective than to test students and remove all doubt. vide a check on the information that emerges from And even when they do participate in assessments, national assessments. And international assessments governments sometimes decline to release the learn- can be powerful tools politically: because country ing results to the public, as happened with the 1995 leaders are concerned with national productivity and TIMSS in Mexico.65 Finally, if assessments are poorly competitiveness, international benchmarking can designed or inappropriately made into high-stakes 18 | World Development Report 2018 FIGURE O.13 Low-performing countries don’t face sharp tradeo s between learning and other education outp tests, administrators or educators may have an Figure O.13 Low-performing countries don’t incentive to cheat on them, rendering the assessment face sharp trade-offs between learning and results worthless as a guide to policy. other education outputs Measurement doesn’t need to detract from broader education objectives—it can even support them (e.g., creativity, citizenship) A stronger emphasis on measurable learning High-performing country B doesn’t mean that other education outcomes don’t Other outputs matter. Formal education and other opportunities for learning have many goals, only some of which are captured by the usual assessments of literacy, numeracy, and reasoning. Educators also aspire to High-performing help learners develop higher-order cognitive skills, country A including some (like creativity) that are hard to capture through assessments. Success in life also Low-performing country C depends on socioemotional and noncognitive skills— such as persistence, resilience, and teamwork—that a good education helps individuals develop. Education Measured learning systems often have other goals as well: they want to Source: WDR 2018 team. endow students with citizenship skills, encourage civic-minded values, and promote social cohesion. These are widely shared goals of education, and it is by high-performing countries on the education understandable that people will ask whether, espe- frontier. Economists use the concept of the produc- cially in education systems that are already over- tion possibilities frontier to understand how pro- burdened, increasing the emphasis on measurable ducers—or in this case countries—make trade-offs learning will crowd out these other goals. between the production of different goods. This In fact, a focus on learning—and on the educa- idea encapsulates the debates on education policy tional quality that drives it—is more likely to “crowd in OECD countries on the learning frontier (figure in” these other desirable outcomes. Conditions that O.13). For example, in recent years many stakehold- allow children to spend two or three years in school ers in Korea have argued that their high-performing without learning to read a single word, or to reach education system places too much emphasis on test the end of primary school without learning to do scores (called “measured learning” in figure O.13) and two-digit subtraction, are not conducive to reaching not enough on creativity and certain socioemotional the higher goals of education. Schools that cannot skills such as teamwork (“other outputs”). Implicitly, equip youth with relevant job skills usually will not this Korean debate is about whether to try to move prepare them to launch new companies or analyze up and to the left on the frontier—that is, from A great works of literature either. If students cannot toward B. But in the low-learning trap, represented focus because of deprivation, if teachers lack the by “low-performing country C” in the figure, there is pedagogical skills and motivation to engage students, so much slack and such a weak focus on outcomes if materials meant for the classroom never reach it that this OECD-driven debate is not relevant. Coun- because of poor management, and if the system as a try C has an opportunity to improve on both mea- whole is unmoored from the needs of society—well, is sured learning and other education outputs at the it really plausible to believe that students are develop- same time. An experiment in Andhra Pradesh, India, ing higher-order thinking skills like problem-solving that rewarded teachers for gains in measured learn- and creativity? It is more likely that these conditions ing in math and language led to more learning not undermine the quest for higher goals—and that, con- just in those subjects, but also in science and social versely, improving the learning focus would acceler- studies—even though there were no rewards for ate progress toward those goals as well. the latter.66 This outcome makes sense—after all, lit- Paradoxically, lower-performing countries proba- eracy and numeracy are gateways to education more bly do not face the same sharp trade-offs encountered generally. Overview | 19 Act on evidence—to make schools work for four times as much in Israel as it did in Kenya—and it all learners has had no impact in some contexts.72 In the words of Measurement of learning shortfalls two commentators on this literature: “Knowing ‘what doesn’t provide clear guidance on works’ in the sense of the treatment effect on the trial how to remedy them. Fortunately, population is of limited value without understanding there is now a lot of experience the political and institutional environment in which on ways to improve learning out- it is set.”73 comes at the student, classroom, The next section tackles the question of that and school levels. Cognitive neuro­ broader environment, but in the meantime we first Policy science has evolved dramatically address how to use this evidence most effectively. response 2: in the last two decades, providing There are four main considerations. Act on insights on how children learn.67 First, more important than the individual results evidence This work has revealed how import- from individual studies are the principles of how ant the first several years of life are to a child’s brain and why programs work. In economic terms, “princi- development.68 At the same time, schools and systems ples” correspond to models of behavior that can then around the world have innovated in many ways: by help guide broader sets of approaches to addressing deploying novel approaches to pedagogy, using new problems. Three types of models can prove especially technologies to enhance teaching and learning in insightful: straightforward models in which actors classrooms, or increasing the accountability, and some- maximize their welfare subject to the constraints times autonomy, of various actors in the system. The they face; principal-agent models that incorporate number of systematic evaluations of whether these multiple actors with different goals and perhaps dif- interventions have improved learning has increased ferent information; and behavioral models that factor more than 10-fold, from just 19 in 2000 to 299 in 2016.69 in mental models and social norms. Many interventions have succeeded in improving Second, a gap between what the evidence suggests learning outcomes. The learning gains from effec- may be effective and what is done in practice points tive interventions translate into additional years of to a potential entry point for action. Understanding schooling, higher earnings, and lower poverty. For a why gaps open up helps guide how to address them. group of stunted Jamaican children 9–24 months old, For example, when different actors face different a program to improve cognitive and socioemotional information, or some actors lack information, this development led to much better outcomes 20 years suggests drawing from approaches that show how later—lower crime rates, better mental health, and information can be disseminated and used better. earnings that were 25 percent higher than those of Gaps point to which types of principles should drive nonparticipants.70 Programs to improve pedagogy context-specific innovation. have had an impact greater than the equivalent of an Third, evidence tends to accumulate where it extra half a year of business-as-usual schooling and is easiest to generate, not necessarily where action an 8 percent increase in the present discounted value would make the most difference, so policies focused of lifetime earnings.71 So while tackling the learning only on that evidence might be misguided.74 Though crisis is hard, the fact that there are interventions that the scope of the accumulated evidence in education is improve learning suggests ways forward. broad, just because an approach hasn’t been evaluated This evidence base does not allow us to identify doesn’t mean it lacks potential. Context-specific inno- what works in all contexts because there are no vation may mean trying things that have not been global solutions in education. Improving learning in tried elsewhere. a particular setting will never be as simple as taking Fourth, a focus on underlying principles high- a successful program from one country or region and lights that the problem can’t be solved by one decision implementing it elsewhere. Randomized controlled maker simply prescribing an increase in the quantity, trials and other approaches to evaluate impact place or even the quality, of one or more inputs. Many of a premium on carefully isolating the causal impact the inputs in learning are the result of choices made of an intervention. But such approaches may ignore by the various actors—choices made in reaction to important interactions with underlying factors that the actual and anticipated choices of other actors. affect whether an intervention makes a difference— For example, teachers respond to incentives to attend factors that may not be at play when replicating the school and to improve student outcomes, even intervention in a new context. For example, increas- though the nature of the response varies across con- ing class size by 10 students reduced test scores by texts.75 Likewise, students and parents make choices 20 | World Development Report 2018 Figure O.14 It’s more complicated than it looks: People act in reaction to the choices of others throughout the system Civil so ns ciet cia y liti o rg Po an iz at io ns r Pe c to s Le er er ar se ch s/c a te ne a om Te rs Priv mun ities LEARNING Sch B u re ts oo pu y lm a uc an iar in ag o ol dic rat em ent S ch s In te Ju rn at ion o rs al a act c to r s O ther Source: WDR 2018 team. responding to other decisions. In India and Zambia, There are three key entry points to addressing learner government grants to schools led parents to reduce preparation: their own investments in their children’s schooling.76 All things considered, a more complete character- Set children on high-development trajectories through •  ization of the learning framework might be closer early childhood nutrition, stimulation, and care. Three to the one illustrated in figure O.14: learning how approaches stand out from successful experiences. to improve outcomes by intervening at the student, First, target mothers and their babies with health classroom, and school l ­ evels involves illuminating the and nutrition interventions during the first 1,000 various arrows. days to reduce malnutrition and foster physiolog- Putting all this together sheds light on three sets ical development. Second, increase the frequency of promising entry points: prepared learners, effec- and quality of stimulation and opportunities for tive teaching, and school-level interventions that learning at home (starting from birth) to improve actually affect the teaching and learning process. language and motor development, as well as to Each of these priority areas is founded on evidence cultivate early cognitive and socioemotional skills. from multiple contexts showing that it can make a Third, promote day-care centers for very young real difference for learning. children and preschool programs for children 3–6 years old—along with caregiver programs that Prepare children and youth for learning enhance the nurturing and protection of children— Getting learners to school ready and motivated to to improve cognitive and socioemotional skills in learn is a first step to better learning. Without it, other the short run, as well as education and labor market policies and programs will have a minimal effect. outcomes later in life.77 Program quality matters Overview | 21 a lot: center-based programs with poor process For effective teacher training, design it to be individually •  quality (even with relatively good infrastructure, targeted and repeated, with follow-up coaching—often caregiver training, and caregiver-children ratios) around a specific pedagogical technique. This approach can actually worsen developmental outcomes.78 contrasts starkly with much of today’s profes- •  Lower the cost of schooling to get children into school, but sional development for teachers across a range of then use other tools to boost motivation and effort because countries. In the United States, a team of teacher cost-reducing interventions don’t usually lead to learning training experts characterized professional devel- on their own.79 To improve learning, demand-side pro- opment there as “episodic, myopic, and often mean- grams need to increase a student’s effort or capacity ingless.”90 In Sub-Saharan Africa, teacher training is to learn. School-provided meals, for example, have often too short to be effective and too low in quality had positive effects on access—and also on learning to make a difference.91 By contrast, programs in in places where children have limited access to food Africa and South Asia that provided long-term at home.80 Targeted cash transfers have led to more coaching led to sizable learning gains.92 learning when they have incentivized performance To keep learners from falling behind to the point where •  itself81 or were marketed in a way that induces more they cannot catch up, target teaching to the level of the effort, such as in Cambodia.82 Some information student. Over the course of several grades, often interventions have motivated efforts as well.83 only a fraction of learners progress at grade level, •  To make up for the fact that so many youth lack skills with most falling behind and some learning almost when leaving basic education, provide remediation before nothing. This is partly because teachers teach to the further education and training.84 Remediation in school most advanced students in the class, as documented is a first best approach. After school, the more suc- from Australia to Sweden to the United States,93 or cessful programs share two main features. First, because the curriculum is too ambitious but teach- they provide bridging courses in real-life settings, ers are required to teach it.94 Effective strategies to which allows learners with very low foundational target teaching to the level of the student include skills to build these in the workplace.85 Second, accel- using community teachers to provide remedial les- erated, flexible pathways—not sequential courses sons to the lowest performers, reorganizing classes over multiple semesters—are associated with greater by ability, or using technology to adapt lessons to student retention and ultimate certification.86 individual student needs.95 Use pecuniary and nonpecuniary incentives to improve •  Make teaching more effective the motivation of teachers, ensuring that the incentiv- Effective teaching depends on teachers’ skills and ized actions are within teachers’ capacity. Education motivation, and yet many systems do not take systems typically neither reward teachers for per- them seriously. Teacher salaries are the largest forming well nor penalize them for performing single budget item in education systems, consuming poorly. Incentives are most likely to be effective three-quarters of the budget at the primary level in at improving outcomes when there are straight- developing countries. Yet many systems struggle forward actions that teachers can take to improve to attract strong candidates into teaching and to learning—such as increasing attendance when provide a solid foundation of subject or pedagogical absenteeism is the constraint. But incentives do not knowledge before they start teaching. As a result, new need to be high stakes (or financial) to affect behav- teachers often find themselves in classrooms with ior. In Mexico and Punjab, Pakistan, simply provid- little mastery of the content they are to teach.87 Once ing diagnostic information to parents and schools teachers are in place, the professional development about the schools’ relative performance improved they receive is often inconsistent and overly theo- learning outcomes.96 retical. In some countries, the cost of this training is enormous, reaching $2.5 billion a year in the United Focus everything else on teaching and States.88 Moreover, education systems often have few learning effective mechanisms in place to mentor, support, School inputs, management, and governance must and motivate teachers—even though teachers’ skills benefit the learner-teacher relationship if they are do nothing for learning unless teachers choose to to improve learning—but many do not. Debates on apply them in the classroom.89 Fortunately, teachers’ improving education outcomes frequently revolve skills and motivation can be strengthened, leading around increasing inputs, such as textbooks, technol- to greater effort and more learning, with three main ogy, or school infrastructure. But too often the question promising principles emerging: of why these inputs might actually improve learning is 22 | World Development Report 2018 overlooked. The evidence on successful use of inputs it is high), and when a range of stakeholders (not and management suggests three main principles: just parents) are brought together in ways that lead to action. In Indonesia, school grants improved Provide additional  inputs, including new technologies, •  learning when links between the school and the in ways that complement rather than substitute for village council—a center of local authority—were teachers.97 A computer-assisted learning program strengthened.109 in Gujarat, India, improved learning when it added to teaching and learning time, especially for the The most effective systems—in terms of learn- poorest-performing students.98 A Kenyan program ing—are those that have narrowed gaps between that provided public school teachers with tablets evidence and practice. On learner preparation, for to support instruction increased the reading per- example, East Asian countries such as Korea and Sin- formance of their students.99 But simply providing gapore have achieved high levels of children ready to desktop computers to classrooms in Colombia— learn. Stunting rates among preschool-age children where they were not well integrated with the cur- are low, and children are motivated and supported by riculum—had no impact on learning.100 Even more their families. To promote effective teaching, Finland traditional inputs—such as books—often fail to and Singapore attract some of the most highly skilled affect teaching and learning when they aren’t actu- graduates from tertiary education into teaching and ally deployed in classrooms, or if the content is too provide them with effective professional develop- advanced for the students.101 ment opportunities and sustained support. Ensure that new information and communication tech- •  nology is really implementable in the current systems. Align actors—to make the whole system Interventions that incorporate information and work for learning communication technology have some of the big- Working at scale is not just “scaling gest impacts on learning.102 But for every highly up.” The concept of scaling up in effective program—such as a dynamic computer- education implies taking interven- assisted learning program for secondary school tions that have been shown to be students in Delhi that increased math and language effective on a pilot or experimental scores more than the vast majority of other learning scale and replicating them across interventions tested in India or elsewhere103—there hundreds or thousands of schools. are programs such as the One Laptop Per Child pro- However, this approach often fails Policy grams in Peru and Uruguay, which evaluations sug- because the key actors are human response 3: gested had no impact on student reading or math beings, operating with human Align actors ability.104 Technologies ill-adapted to their settings aspirations and limitations in a often fail to reach the classroom or to be used if they politically charged arena. Real-world complications reach it.105 can undermine well-designed programs, especially •  Focus school management and governance reforms when new, systemwide forces come into play. When on improving teacher-learner interaction. Training the Cambodian government tried to scale up early principals in how to improve that interaction—by child development centers and preschools—programs providing feedback to teachers on lesson plans, that had worked in some parts of the country when action plans to improve student performance, implemented by nongovernmental organizations and classroom behavior—has led to a large impact (NGOs)—low demand from parents and low-quality on student learning.106 In countries ranging from services led to no impacts on child development, Brazil and India to Sweden, the United Kingdom, and even slowed it for some.110 When the Kenyan and the United States, the management capacity of government tried to lower student-teacher ratios by school principals significantly and robustly relates hiring contract teachers—an intervention that had to student performance—even after controlling for improved student outcomes when implemented by a variety of student and school characteristics.107 an NGO—the results were negligible because of both Involving communities, parents, and school actors implementation constraints and political economy in ways that promote local oversight and account- factors.111 And when the Indonesian government ability for service delivery can improve outcomes.108 tried to increase teacher capacity by nearly doubling But community monitoring tends to have more the salaries of certified teachers, political pressures impact when it covers things that parents can watered down the certification process and left only easily observe (such as teacher absenteeism when the pay increase in place. The result was much larger Overview | 23 budget outlays on salaries, but no increase in teach- the absence of good information on learning prevents ers’ skills or student learning.112 stakeholders from judging system performance, The lesson, then, is that better interventions at the designing the appropriate policies, and holding poli- school and student levels will sustainably improve ticians and bureaucrats to account. Thus improving learning only if countries tackle the stubborn system- learning metrics is crucial for drawing attention to level technical and political barriers to change. Tech- problems and building the will for action. In Tanzania nical barriers include the complexity of the system, in the early 2010s, poor results on school-leaving exam- the large number of actors, the interdependence of inations—along with well-publicized results from reforms, and the slow pace of change in education citizen-led learning assessments and surveys showing systems. Political barriers include the competing poor service delivery in schools—motivated policy interests of different players and the difficulty of makers to launch ambitious reforms. In Germany, the moving out of a low-quality equilibrium, especially shock of mediocre results on the first PISA in 2000 led in low-trust environments where risks predominate. to reforms that improved both learning and equity. All of these barriers pull actors away from learning, Efforts in this area need to go beyond just measur- as discussed earlier. Systems that surmount these ing learning; they should track its determinants as barriers and align actors toward learning can achieve well. Understanding these determinants can enable remarkable learning outcomes. Shanghai provided reforms to grapple with the deeper causes, if there proof when it topped the 2012 PISA rankings, in part is a systemwide commitment to improving learning. thanks to policies that ensured that every classroom Take the issue of learner preparedness. When indica- had a prepared, supported, and motivated teacher.113 tors reveal that poorer children already lag far behind To shift the system toward learning, technically by the time they start primary school, this finding and politically, reformers can use three sets of tools: can build political will not only to expand preschool education in low-income areas, but also to combat • Information and metrics. Better information and stunting and educate parents about early stimulation metrics can promote learning in two ways: by of children. When indicators show that many teach- catalyzing reforms and by serving as indicators of ers lack a strong command of what their students whether reforms are working to improve learning are meant to learn, this finding can spark efforts to with equity. Thus they can improve both the politi- improve the quality of teacher education.114 cal and technical alignment of the system. Of course, information and metrics can also be • Coalitions and incentives. Good information will have misleading, irrelevant, or politically unsustainable, a payoff only if there is enough support for prior- so they need to be designed and used wisely. Metrics itizing learning. Politics is often the problem, and may fail to capture important dimensions of the politics must be part of the solution. This requires outcomes the education system is trying to promote. forming coalitions to advocate for broad-based For example, the Millennium Development Goal of learning and skills and to rebalance the political universal primary education by 2015 embodied a cru- incentives. cial goal—equitable access—but it did not represent • Innovation and agility. Schools and societies have what many assumed it did: universal acquisition of achieved high levels of equitable learning in a foundational literacy and numeracy, let alone other variety of ways. Figuring out what approaches will life skills. Another risk is of distorting good metrics work in a given context requires innovation and by putting high stakes on them, if potential benefi- adaptation. This means using evidence to identify ciaries can game the indicators. Thus systems will where to start and then using metrics to iterate need different measures for different purposes.115 with feedback loops. Even if they are technically sound, metrics may prove politically unsustainable if they highlight too All of these tools will be most effective when sup- many problems and do not provide any reason for ported by strong implementation capacity within hope. One way to address this problem is to focus not government. on levels of learning, which may be very low, but on progress over time. Information and metrics Better information and measurement—starting with Coalitions and incentives learning metrics—are critical to creating political Mobilizing everyone who has a stake in learning space for innovation and then using that space to has been an important strategy in efforts to improve achieve continuous improvement. As emphasized, learning. Many countries have used wide-ranging 24 | World Development Report 2018 consultations that have tried to bring in all interest Innovation and agility groups to build support for proposed changes in edu- To develop effective learning approaches that fit their cation policy. Malaysia used a “lab” model to bring contexts, education systems need to encourage inno- together coalitions of stakeholders and involve them vation and adaptation. In many education systems, in all stages of reform, from design to implementa- schools and other education institutions regularly tion.116 Mobilizing citizens through regular informa- adapt to changing circumstances. Through these tion and communication campaigns can also be an adaptations, innovative solutions to education chal- important strategy. In Peru, reformers in the govern- lenges often emerge. Exploring the well-performing ment used information on poor learning outcomes parts of any education system can reveal technically and performance of the education system to mobilize and politically feasible approaches to the problems public support for reforms to strengthen teacher systems face in improving learning. For example, in accountability. That information also catalyzed action Misiones province in Argentina high student dropout by the business community, which funded a cam- rates were widespread, but some schools seemed to paign highlighting the importance of quality educa- buck the trend. A closer look at these “positive devi- tion for economic growth. In parts of Peru, parents ants” revealed very different relationships between used this entry point to protest teacher strikes that teachers and parents. When other schools adopted had disrupted schooling.117 Another tool for building the more constructive approach to parent-teacher coalitions is to bundle reforms, so that each actor relations used by the successful schools, their drop- achieves one of its top priorities. For example, a com- out rates fell significantly.122 Burundi, while recover- mitment to modernize vocational training—a reform ing from a civil war, used an adaptive approach to find that could help employers immediately—could buy the right way to get textbooks to schools. It reduced their support for broader education reforms. delivery times from over a year to 60 days—then rep- Where feasible, a negotiated and gradual approach licated that approach in other areas.123 to reform can provide a more promising alternative Incentives are important in determining whether to direct confrontation. When system actors agree to systems innovate and adopt emerging solutions at collaborate and build trust around shared goals, the scale. Systems that are closed, that limit the autonomy chances of successful reform are likely to be higher. of teachers and schools, and that judge performance by In Chile, successive negotiations between the govern- the extent of compliance with rules governing resource ment and the teachers’ union built broad support for a use often provide little room for innovation. By contrast, series of reforms that adjusted the working conditions more open systems that pay more attention to overall of teachers to improve their overall welfare, while outcomes and reward progress in raising outcomes are linking pay and career development more closely to more likely to see greater innovation and the diffusion performance.118 One approach used by several coun- of new approaches across the education system.124 tries has been to compensate actors who might lose To make a difference at the system level, such out from reforms. In other cases, dual-track reforms innovations needs to be packaged with good metrics have been introduced to phase in changes in a way and with system-level coalitions for learning. With- that protects incumbent actors from their effects—for out both, any improvements from innovation are example, in Peru and the District of Columbia in the likely to prove short-lived or limited to local areas. But United States, pay-for-performance schemes were with such support, a virtuous cycle becomes possible initially voluntary.119 as systems follow these steps: Building strong partnerships between schools and their communities is also important for sustaining Set learning as a clearly articulated goal and mea- •  reforms. Where political and bureaucratic incen- sure it. tives for reform are weak, action at the local level Build a coalition for learning that gives the political •  can act as a substitute. In South Africa, the political space for innovation and experimentation. and economic context constrains efforts to improve Innovate and test approaches that seem the most •  education performance. Yet progress was made in promising for the given context, drawing inspira- improving outcomes at the local level through strong tion from the evidence base and focusing on areas partnerships between parents and schools.120 Even that promise the biggest improvements over cur- where broader incentives exist to improve learning, rent practice. community engagement at the local level is import- Use the measure of learning, along with the other •  ant and can complement national or subnational metrics of delivery, as a gauge of whether the change efforts.121 approach is working. Overview | 25 FIGURE O.15 Coherence and alignment towards learning Figure O.15 Coherence and alignment toward learning CiviC l iv il so ians orgS o anic ietciet t ic P oli z at y Oy ion r s gs . Pe rs Le e ar er r ch to s/c ec ne a om Te te s rs mun Priva ities LEARNING Sch Bure iary oo ts a uc pu lm dic na in l rat a Ju ge oo s me h nt Sc In te rn s at i or on al a r act c to r s Ot he Source: WDR 2018 team. Build on what works, and scale back what doesn’t, •  teaching-learning relationship, or because the system to deliver short-term results that strengthen the doesn’t prioritize learning for disadvantaged children long-term resolve of the coalition for learning. and youth. More financing for business as usual will Repeat. •  therefore just lead to the usual outcomes. But where countries seriously tackle the barriers to learning for The payoff to doing what needs to be done is a all, spending on education is a critical investment for system in which the elements are coherent with each development, especially for those countries where other and everything aligns with learning (figure O.15). overall spending is currently low, as recent major Increased financing can support this learning- studies of global education have emphasized.126 More for-all equilibrium, if the various key actors behave children staying in school longer and learning while in ways that show learning matters to them. This is there will undoubtedly require more public financing a big “if” because higher levels of public spending are for education. An injection of financing—either from not associated statistically with higher completion or domestic or international sources—can help coun- even enrollment rates in countries with weak gover- tries escape the low-learning trap, if they are willing nance.125 Ensuring that students learn is even more to take the other necessary steps laid out here. challenging, and so there is little correlation between spending and learning after accounting for national Implications for external actors income. It is easy to see the reason for this because External actors can reinforce these strategies for of the many ways in which financing can leak out— opening the political and technical space for learn- whether because money never reaches the school, ing. In the realm of information and metrics, for or because it pays for inputs that don’t affect the example, international actors can fund participation 26 | World Development Report 2018 in regional learning assessments (such as PASEC But waiting out the learning crisis isn’t a winning in West Africa or LLECE in Latin America) or global strategy. Even though national income and learning learning assessments (such as PISA or TIMSS) to are somewhat correlated at lower levels of develop- spotlight challenges and catalyze domestic efforts ment, higher incomes do not invariably lead to better for reform. External actors can also develop tools learning outcomes. And to the extent that develop- for tracking the proximate determinants of learning ment does bring better learning and skills, it is partly to aid in feedback loops. Domestic financing usu- because development has been accompanied by a ally makes up the bulk of education financing, so a willingness to tackle the political impasses and gov- high-leverage entry point for international actors is ernance challenges that hamper learning. Ultimately, to fund better information that will make domestic then, those challenges are not avoidable. Furthermore, spending more effective. In the realm of innovation there’s no need to wait for learning. At every level of and experimentation, external funders such as the income, there are countries that not only score better World Bank can provide results-based financing that than others on international assessments, but also— gives countries more room to innovate and iterate and more important—show from the quality of their their way to achieving better outcomes. education systems and their policy making that they are committed to learning. Learning to realize The future of work will place a premium on learn- ing. Rapid technological change has led to major education’s promise shifts in the nature of work, leading some to declare By showing that learning really matters to them, this a new era—the Second Machine Age or the Fourth countries can realize education’s full promise. Beyond Industrial Revolution. In the extreme versions of this being a basic human right, education—done right— vision, all but a few jobs could disappear, decreasing improves social outcomes in many spheres of life. the value of skills for most people. But the seismic For individuals and families, education boosts human changes predicted have yet to permeate the high- capital, improves economic opportunities, promotes income countries, let alone the low- and middle- health, and expands the ability to make effective income ones. More important, no matter how the choices. For societies, education expands economic demand for skills changes in the future, people opportunities, promotes social mobility, and makes will require a solid foundation of basic skills and institutions function more effectively. In measuring knowledge. If anything, rapid change will increase these benefits, research has only recently focused on the returns to learning how to learn, which requires the distinction between schooling and learning. But foundational skills that allow individuals to size the evidence confirms the intuition that these benefits up new situations, adapt their thinking, and know often depend on the skills that students acquire, not where to go for information and how to make sense just the number of years in the classroom. Economies of it. with higher skills grow faster than those with school- ing but mediocre skills; higher literacy predicts better *** financial knowledge and better health, beyond the Countries have already made a tremendous start by effects of schooling; and poor children are more likely getting so many children and youth into school. Now to rise in the income distribution when they grow up it’s time to realize education’s promise by accelerat- in communities with better learning outcomes. ing learning. A real education—one that encourages Taking learning seriously won’t be easy. It’s hard learning—is a tool for promoting both shared pros- enough to work through the technical challenges of perity and poverty elimination. That type of educa- figuring out what will promote learning at the level of tion will benefit many: children and families whose the student and school in any context, let alone tackle positive schooling experience restores their faith in the political and technical challenges of working at government and society rather than eroding it; youth scale. Many countries struggling with the learning who have skills employers are seeking; teachers who crisis may be tempted to continue with business as can respond to their professional calling rather than usual. After all, they may reason, development will to political demands; adult workers who have learned eventually improve learning outcomes: as households how to learn, preparing them for unforeseeable eco- escape poverty and schools take advantage of better nomic and social changes; and citizens who have the facilities, more materials, and better-trained teachers, values and reasoning abilities to contribute to civic better learning outcomes should follow. life and social cohesion. Overview | 27 Notes 1. Uwezo (2014). In all countries, the test was administered 32. Black and others (2017). Stunting is defined by the World in English. In Kenya and Tanzania, it was also admin- Health Organization (WHO) as a height-for-age z-score istered in Kiswahili, and the highest score (English or of less than two standard deviations below the median Kiswahili) was used in the assessment of proficiency. of a healthy reference population. English is the language of instruction in Kenya and 33. Paxson and Schady (2007); Schady and others (2015). Uganda. 34. Hanushek (1992); Rockoff (2004). 2. ASER Centre (2017). 35. Bau and Das (2017). 3. WDR 2018 team, using data from the Programme for 36. Bruns and Luque (2015). International Student Assessment (PISA), 2015 (OECD 37. UIS (2006). 2016). 38. Chang and others (2013). 4. WDR 2018 team, using data from the Third Regional 39. Abadzi (2009); EQUIP2 (2010). Comparative and Explanatory Study (TERCE), 2012 40. Bold and others (2017). (UNESCO 2013). 41. Hanushek (1995); Mingat and Tan (1998); Tan and 5. UNESCO (2016). Mingat (1992); Wolf (2004). 6. World Bank (2011). 42. Glewwe and others (2011); Hanushek (1986); Kremer 7. Barro and Lee (2013). (1995). 8. Pritchett (2013). 43. Sabarwal, Evans, and Marshak (2014). 9. Pritchett (2013). 44. Lavinas and Veiga (2013). 10. Gove and Cvelich (2011). 45. Robinson, Lloyd, and Rowe (2008); Waters, Marzano, 11. Crouch (2006). and McNulty (2003). 12. Castillo and others (2011). 46. Bloom and others (2015). Management areas include 13. ASER Pakistan (2015a, 2015b). operations, monitoring, target setting, and people 14. Southern and Eastern Africa Consortium for Monitor- management. ing Educational Quality (SACMEQ) results for grade 6 47. Bruns, Filmer, and Patrinos (2011); Orazem, Glewwe, and students in 15 countries in 2007 (Hungi and others Patrinos (2007); World Bank (2003). 2010). 48. Data extracted from U.S. Agency for International 15. Programme d’Analyse des Systèmes Éducatifs de la Development’s Demographic and Health Survey (DHS) Confemen (PASEC) results for grade 6 students in 10 StatCompiler, http://www.statcompiler.com/en/. francophone countries in 2014 (PASEC 2015). 49. Park (2016). 16. ASER Centre (2017). 50. Todd and Mason (2005). 17. RTI International (2009). 51. Chisholm and Leyendecker (2008). 18. World Bank (2016b). 52. World Bank (2003). 19. Muralidharan and Zieleniak (2013). 53. Andrews, Pritchett, and Woolcock (2017). 20. Spaull and Kotze (2015). 54. Grindle (2004). 21. Singh (2015). 55. Evans and Yuan (2017). 22. Minimum proficiency is defined as one standard devi- 56. The team thanks Kai-Ming Cheng for suggesting this ation below the mean of the harmonized assessment formulation. scores. 57. Dweck (2008). 23. These numbers are based on analysis of the data in 58. Save the Children (2013). “A Global Data Set on Education Quality” (2017), a data 59. Guilfoyle (2006). set made available to the WDR 2018 team by Nadir Alti- 60. Jacob (2005). nok, Noam Angrist, and Harry Anthony Patrinos. These 61. Fausset (2014). averages do not include China or India because of lack 62. UIS (2016). of data. 63. OECD (2011). 24. UNESCO (2016). 64. Jacob (2007). 25. UIS and EFA (2015). 65. Solano-Flores, Contreras-Niño, and Backhoff Escudero 26. Banerjee, Jacob, and Kremer (2000); Hanushek and (2005). Woessmann (2008); Rivkin, Hanushek, and Kain (2005). 66. Muralidharan and Sundararaman (2011). 27. Alderman, Orazem, and Paterno (2001); Andrabi, Das, 67. De Smedt (2014); Insel and Landis (2013); Kuhl (2010). and Khwaja (2008); Farah (1996); Kingdon (1996); 68. Dua and others (2016). Orazem (2000); Tooley and Dixon (2007). 69. Evans and Popova (2016). 28. Hanushek, Lavy, and Hitomi (2008). 70. Gertler and others (2014). 29. STEP surveys (World Bank 2014). 71. Calculations carried out for WDR 2018. See Evans and 30. Lupien and others (2000); McCoy and others (2016); Yuan (2017). Walker and others (2007). 72. Pritchett and Sandefur (2013). 31. Coe and Lubach (2007); Garner and others (2012); 73. Deaton and Cartwright (2016). Nelson (2016). 74. Romer (2015). 28 | World Development Report 2018 75. Duflo, Hanna, and Ryan (2012); Muralidharan and 112. Chang and others (2013); de Ree and others (2015). Sundararaman (2011). 113. Liang, Kidwai, and Zhang (2016). 76. Das and others (2013). 114. For example, in Mozambique, after the World Bank’s 77. The evidence is from countries ranging from the Service Delivery Indicators revealed very low levels of United States to Argentina, Bangladesh, China, and teacher knowledge and very high levels of absentee- Uganda, among others (Berlinski, Galiani, and Gertler ism—results that were picked up by the local media— 2008; Engle and others 2011). the government launched a program (ultimately sup- 78. Berlinski and Schady (2015); Bernal and others (2016); ported through a loan from the World Bank) to address Grantham-McGregor and others (2014). these issues. 79. Baird and others (2014); Fiszbein and Schady (2009); 115. Neal (2013). Morgan, Petrosino, and Fronius (2012). 116. World Bank (2017). 80. Snilstveit and others (2016). 117. Bruns and Luque (2015). 81. Blimpo (2014); Kremer, Miguel, and Thornton (2009). 118. Mizala and Schneider (2014); Wales, Ali, and Nicolai Direct financial incentives have been less successful in (2014). high-income countries (Fryer 2011), although alternate 119. Birnbaum (2010); Bruns and Luque (2015). designs that deliver incentives immediately after the 120. Levy and others (2016). test have worked (Levitt and others 2016). 121. Mansuri and Rao (2013). 82. Barrera-Osorio and Filmer (2013). 122. Green (2016); Pascale, Sternin, and Sternin (2010). 83. Avitabile and de Hoyos (2015); Nguyen (2008). 123. Campos, Randrianarivelo, and Winning (2015). 84. ILO (2015). 124. Andrews, Pritchett, and Woolcock (2013). 85. Bragg (2014). 125. Rajkumar and Swaroop (2008); Suryadarma (2012). 86. Calcagno and Long (2008); Martorell and McFarlin Jr. 126. See, in particular, the report of the Education Commis- (2011); Scott-Clayton and Rodriguez (2014). sion (2016), which emphasizes the important role of 87. Tandon and Fukao (2015); World Bank (2013, 2016a). finance in complementing reforms. 88. Layton (2015). 89. Bruns and Luque (2015); Mulkeen (2010). 90. Darling-Hammond and others (2009). 91. Lauwerier and Akkari (2015). References 92. Banerjee and others (2007); Conn (2017). Abadzi, Helen. 2009. “Instructional Time Loss in Developing 93. Abadzi and Llambiri (2011); Ciaccio (2004); Leder (1987). Countries: Concepts, Measurement, and Implications.” 94. Banerjee and others (2016); Pritchett and Beatty (2015). World Bank Research Observer 24 (2): 267–90. 95. Banerjee and others (2007); Duflo, Dupas, and Kremer Abadzi, Helen, and Stavri Llambiri. 2011. “Selective Teacher (2011); Kiessel and Duflo (2014); Muralidharan, Singh, Attention in Lower-Income Countries: A Phenomenon and Ganimian (2016). Linked to Dropout and Illiteracy?” Prospects 41 (4): 96. Andrabi, Das, and Khwaja (2015); de Hoyos, Garcia- 491–506. Moreno, and Patrinos (2017). Alderman, Harold, Peter F. Orazem, and Elizabeth M. Paterno. 97. Snilstveit and others (2016). 2001. “School Quality, School Cost, and the Public/Private 98. Linden (2008). School Choices of Low-Income Households in Pakistan.” 99. Piper and others (2015). Journal of Human Resources 36 (2): 304–26. 100. Barrera-Osorio and Linden (2009). Andrabi, Tahir, Jishnu Das, and Asim Ijaz Khwaja. 2008. 101. Glewwe, Kremer, and Moulin (2009); Sabarwal, Evans, “A Dime a Day: The Possibilities and Limits of Private and Marshak (2014). Schooling in Pakistan.” Comparative Education Review 52 102. McEwan (2015). (3): 329–55. 103. Muralidharan, Singh, and Ganimian (2016). —— ——. 2015. “Report Cards: The Impact of Providing School 104. Cristia and others (2012); De Melo, Machado, and and Child Test Scores on Educational Markets.” Policy Miranda (2014). For Uruguay, the evaluation covers Research Working Paper 7226, World Bank, Washing- math and reading impacts in the early years of the ton, DC. program, when its main objective was to provide Andrews, Matt, Lant Pritchett, and Michael Woolcock. 2013. equipment and connectivity for schools; the program “Escaping Capability Traps through Problem Driven Iter- evolved since then to add ICT training for teachers and ative Adaptation (PDIA).” World Development 51: 234–44. adaptive educational technology, and new evaluations —— ——. 2017. Building State Capability: Evidence, Analysis, Action. are expected to be published in late 2017. New York: Oxford University Press. 105. Lavinas and Veiga (2013). ASER Centre. 2017. Annual Status of Education Report (Rural) 106. Fryer (2017). 2016. New Delhi: ASER Centre. http://img.asercentre.org 107. Bloom and others (2015). /docs/Publications/ASER%20Reports/ASER%202016 108. Bruns, Filmer, and Patrinos (2011). /aser_2016.pdf. 109. Pradhan and others (2014). ASER Pakistan. 2015a. “Annual Status of Education Report: 110. Bouguen and others (2013). ASER Pakistan 2015 National (Rural).” Lahore, Pakistan: 111. Bold and others (2013). South Asian Forum for Education Development. Overview | 29 ASER Pakistan. 2015b. “Annual Status of Education Report: .washingtonpost.com/wp-dyn/content/article/2010/09/10 ASER Pakistan 2015 National (Urban).” Lahore, Pakistan: /AR2010091006604.html. South Asian Forum for Education Development. Black, Maureen M., Susan P. Walker, Lia C. H. Fernald, Avitabile, Ciro, and Rafael E. de Hoyos. 2015. “The Heteroge- Christopher T. Andersen, Ann M. DiGirolamo, Chunling neous Effect of Information on Student Performance: Lu, Dana C. McCoy, et al. 2017. “Early Childhood Develop- Evidence from a Randomized Control Trial in Mexico.” ment Coming of Age: Science through the Life Course.” Policy Research Working Paper 7422, World Bank, Wash- Lancet 389 (10064): 77–90. ington, DC. Blimpo, Moussa P. 2014. “Team Incentives for Education in Baird, Sarah Jane, Francisco H. G. Ferreira, Berk Özler, and Developing Countries: A Randomized Field Experiment Michael Woolcock. 2014. “Conditional, Unconditional in Benin.” American Economic Journal: Applied Economics and Everything in Between: A Systematic Review of the 6 (4): 90–109. Effects of Cash Transfer Programmes on Schooling Out- Bloom, Nicholas, Renata Lemos, Raffaella Sadun, Daniela comes.” Journal of Development Effectiveness 6 (1): 1–43. Scur, and John Van Reenen. 2014. “JEEA-FBBVA Lecture Banerjee, Abhijit Vinayak, Rukmini Banerji, James Berry, 2013: The New Empirical Economics of Management.” Esther Duflo, Harini Kannan, Shobhini Mukherji, Marc Journal of the European Economic Association 12 (4): 835–76. Shotland, et al. 2016. “Mainstreaming an Effective Inter- Bloom, Nicholas, Renata Lemos, Raffaella Sadun, and John vention: Evidence from Randomized Evaluations of Van Reenen. 2015. “Does Management Matter in ‘Teaching at the Right Level’ in India.” NBER Working Schools?” Economic Journal 125 (584): 647–74. Paper 22746, National Bureau of Economic Research, Bold, Tessa, Deon Filmer, Gayle Martin, Ezequiel Molina, Cambridge, MA. Brian Stacy, Christophe Rockmore, Jakob Svensson, Banerjee, Abhijit Vinayak, Shawn Cole, Esther Duflo, and et al. 2017. “What Do Teachers Know and Do? Does It Leigh Linden. 2007. “Remedying Education: Evidence Matter? Evidence from Primary Schools in Africa.” Policy from Two Randomized Experiments in India.” Quarterly Research Working Paper 7956, World Bank, Washing- Journal of Economics 122 (3): 1235–64. ton, DC. Banerjee, Abhijit Vinayak, Suraj Jacob, and Michael Kremer. Bold, Tessa, Mwangi Kimenyi, Germano Mwabu, Alice 2000. “Promoting School Participation in Rural Rajas- Ng’ang’a, and Justin Sandefur. 2013. “Scaling Up What than: Results from Some Prospective Trials.” With Jenny Works: Experimental Evidence on External Validity in Lanjouw and Peter Lanjouw. Working paper, Massachu- Kenyan Education.” Working Paper 321, Center for Global setts Institute of Technology, Cambridge, MA. Development, Washington, DC. Barrera-Osorio, Felipe, and Deon Filmer. 2013. “Incentivizing Bouguen, Adrien, Deon Filmer, Karen Macours, and Sophie Schooling for Learning: Evidence on the Impact of Alter- Naudeau. 2013. “Impact Evaluation of Three Types of native Targeting Approaches.” Policy Research Working Early Childhood Development Interventions in Cambo- Paper 6541, World Bank, Washington, DC. dia.” Policy Research Working Paper 6540, World Bank, Barrera-Osorio, Felipe, and Leigh L. Linden. 2009. “The Use Washington, DC. and Misuse of Computers in Education: Evidence from a Bragg, Debra D. 2014. “Career Pathways in Disparate Indus- Randomized Experiment in Colombia.” Policy Research try Sectors to Serve Underserved Populations.” Paper Working Paper 4836, World Bank, Washington, DC. presented at American Educational Research Association Barro, Robert J., and Jong Wha Lee. 2013. “A New Data Set of conference, Philadelphia, April 5. Educational Attainment in the World, 1950–2010.” Journal Bruns, Barbara, Deon Filmer, and Harry Anthony Patrinos. of Development Economics 104: 184–98. 2011. Making Schools Work: New Evidence on Accountability Bau, Natalie, and Jishnu Das. 2017. “The Misallocation of Pay Reforms. Human Development Perspectives Series. and Productivity in the Public Sector: Evidence from the Washington, DC: World Bank. Labor Market for Teachers.” Policy Research Working Bruns, Barbara, and Javier Luque. 2015. Great Teachers: How to Paper 8050, World Bank, Washington, DC. Raise Student Learning in Latin America and the Caribbean. Berlinski, Samuel, Sebastian Galiani, and Paul J. Gertler. With Soledad De Gregorio, David K. Evans, Marco 2008. “The Effect of Pre-primary Education on Primary Fernández, Martin Moreno, Jessica Rodriguez, Guill- School Performance.” Journal of Public Economics 93 (1–2): ermo Toral, and Noah Yarrow. Latin American Develop- 219–34. ment Forum Series. Washington, DC: World Bank. Berlinski, Samuel, and Norbert R. Schady, eds. 2015. The Calcagno, Juan Carlos, and Bridget Terry Long. 2008. “The Early Years: Child Well-Being and the Role of Public Policy. Impact of Postsecondary Remediation Using a Regres- Development in the Americas Series. Washington, DC: sion Discontinuity Approach: Addressing Endogenous Inter-American Development Bank; New York: Palgrave Sorting and Noncompliance.” NBER Working Paper Macmillan. 14194, National Bureau of Economic Research, Cam- Bernal, Raquel, Orazio Pietro Attanasio, Ximena Peña, and bridge, MA. Marcos Vera-Hernández. 2016. “The Effects of the Tran- Campos, Jose Edgardo, Benjamina Randrianarivelo, and Kay sition from Home-Based Community Nurseries to Child- Winning. 2015. “Escaping the ‘Capability Trap’: Turning Care Centers on Children in Colombia.” Working paper, ‘Small’ Development into ‘Big’ Development.” Interna- Universidad de los Andes, Bogotá, Colombia. tional Public Management Review 16 (1): 99–131. Birnbaum, Michael. 2010. “D.C. Schools Unveil Teacher-Pay Castillo, Melba, Vanesa Castro, José Ramón Laguna, and Bonus Plan.” Washington Post, September 12. http://www Josefina Vijil. 2011. Informe de Resultados: EGMS Nicaragua. 30 | World Development Report 2018 Research Triangle Park, NC: Centro de Investigación y de Ree, Joppe, Karthik Muralidharan, Menno Pradhan, and Acción Educativa Social and RTI International. https:// Halsey Rogers. 2015. “Double for Nothing? Experimental shared.rti.org/content/informe-de-resultados-egma Evidence on the Impact of an Unconditional Teacher -nicaragua. Salary Increase on Student Performance in Indonesia.” Chang, Mae Chu, Sheldon Shaeffer, Samer Al-Samarrai, NBER Working Paper 21806, National Bureau of Eco- Andrew B. Ragatz, Joppe De Ree, and Ritchie Stevenson. nomic Research, Cambridge, MA. 2013. Teacher Reform in Indonesia: The Role of Politics and De Smedt, Bert. 2014. “Advances in the Use of Neuroscience Evidence in Policy Making. Directions in Development: Methods in Research on Learning and Instruction.” Human Development Series. Washington, DC: World Frontline Learning Research 2 (4): 7–14. Bank. Dua, Tarun, Mark Tomlinson, Elizabeth Tablante, Pia Britto, Chisholm, Linda, and Ramon Leyendecker. 2008. “Curricu- Aisha Yousfzai, Bernadette Daelmans, and Gary L. Darm- lum Reform in Post-1990s Sub-Saharan Africa.” Interna- stadt. 2016. “Global Research Priorities to Accelerate tional Journal of Educational Development 28 (2): 195–205. Early Child Development in the Sustainable Develop- Ciaccio, Joseph. 2004. Totally Positive Teaching: A Five-Stage ment Era.” Lancet Global Health 4 (12): e887–e889. Approach to Energizing Students and Teachers. Alexandria, Duflo, Esther, Pascaline Dupas, and Michael Kremer. 2011. VA: ASCD. “Peer Effects, Teacher Incentives, and the Impact of Coe, Christopher L., and Gabrielle R. Lubach. 2007. “Mother- Tracking: Evidence from a Randomized Evaluation in Infant Interactions and the Development of Immunity Kenya.” American Economic Review 101 (5): 1739–74. from Conception through Weaning.” In Psychoneuro­ Duflo, Esther, Rema Hanna, and Stephen P. Ryan. 2012. immunology, edited by Robert Ader, 455–74. Burlington, “Incentives Work: Getting Teachers to Come to School.” MA: Elsevier Academic Press. American Economic Review 102 (4): 1241–78. Conn, Katharine M. 2017. “Identifying Effective Education Dweck, Carol S. 2008. Mindset, the New Psychology of Success: How We Can Learn to Fulfill Our Potential. New York: Ballan- Interventions in Sub-Saharan Africa: A Meta-Analysis tine Books. of Impact Evaluations.” Review of Educational Research Education Commission. 2016. The Learning Generation: Invest- (May 26). http://journals.sagepub.com/doi/abs/10.3102 ing in Education for a Changing World. New York: Inter­ /0034654317712025. national Commission on Financing Global Education Cristia, Julián P., Pablo Ibarrarán, Santiago Cueto, Ana Opportunity. Santiago, and Eugenio Severín. 2012. “Technology and Engle, Patrice L., Lia C. H. Fernald, Harold Alderman, Jere Child Development: Evidence from the One Laptop Per Behrman, Chloe O’Gara, Aisha Yousafzai, Meena Cabral Child Program.” IZA Discussion Paper 6401, Institute for de Mello, et al. 2011. “Strategies for Reducing Inequalities the Study of Labor, Bonn, Germany. and Improving Developmental Outcomes for Young Crouch, Luis. 2006. “Education Sector: Standards, Account- Children in Low-Income and Middle-Income Countries.” ability, and Support.” In A New Social Contract for Peru: An Lancet 378 (9799): 1339–53. Agenda for Improving Education, Health Care, and the Social EQUIP2 (Educational Quality Improvement Program 2). Safety Net, edited by Daniel Cotlear, 71–106. World Bank 2010. “Using Opportunity to Learn and Early Grade Country Study Series. Washington, DC: World Bank. Reading Fluency to Measure School Effectiveness in Darling-Hammond, Linda, Ruth Chung Wei, Alethea Andree, Ethiopia, Guatemala, Honduras, and Nepal.” Working Nikole Richardson, and Stelios Orphanos. 2009. “Profes- paper, Educational Policy, Systems Development, and sional Learning in the Learning Profession: A Status Management, U.S. Agency for International Develop- Report on Teacher Development in the United States and ment, Washington, DC. Abroad.” National Staff Development Council, Dallas. Evans, David K., and Anna Popova. 2016. “What Really Works Das, Jishnu, Stefan Dercon, James Habyarimana, Pramila to Improve Learning in Developing Countries? An Analy- Krishnan, Karthik Muralidharan, and Venkatesh Sunda- sis of Divergent Findings in Systematic Reviews.” World raraman. 2013. “School Inputs, Household Substitution, Bank Research Observer 31 (2): 242–70. and Test Scores.” American Economic Journal: Applied Eco- Evans, David K., and Fei Yuan. 2017. “Economic Returns to nomics 5 (2): 29–57. Interventions That Increase Learning.” Background Deaton, Angus S., and Nancy Cartwright. 2016. “Understand- paper, World Development Report 2018, World Bank, Wash- ing and Misunderstanding Randomized Controlled ington, DC. Trials.” NBER Working Paper 22595, National Bureau of Farah, I. 1996. “Road to Success: Self-Sustaining Primary Economic Research, Cambridge, MA. School Change in Rural Pakistan.” With T. Mehmood, de Hoyos, Rafael E., Vicente A. Garcia-Moreno, and Harry Amna, R. Jaffar, F. Ashams, P. Iqbal, S. Khanam, Z. Shah, Anthony Patrinos. 2017. “The Impact of an Accountability and N. Gul-Mastoi. Institute for Educational Develop- Intervention with Diagnostic Feedback: Evidence from ment, Aga Khan University, Karachi, Pakistan. Mexico.” Economics of Education Review 58: 123–40. Fausset, Richard. 2014. “Trial Opens in Atlanta School Cheat- De Melo, Gioia, Alina Machado, and Alfonso Miranda. 2014. ing Scandal.” New York Times, September 29. https://www “The Impact of a One Laptop Per Child Program on .nytimes.com/2014/09/30/us/racketeering-trial-opens-in Learning: Evidence from Uruguay.” IZA Discussion -altanta-schools-cheating-scandal.html?_r=1. Paper 8489, Institute for the Study of Labor, Bonn, Filmer, Deon. 2016. “Educational Attainment and Enrollment Germany. around the World: An International Database.” World Overview | 31 Bank, Washington, DC. http://go.worldbank.org/3 of Dropout Behavior in Developing Countries.” Journal of GEREWJ0E0. Human Capital 2 (1): 69–105. Fiszbein, Ariel, and Norbert R. Schady. 2009. Conditional Hanushek, Eric A., and Ludger Woessmann. 2008. “The Role Cash Transfers: Reducing Present and Future Poverty. With of Cognitive Skills in Economic Development.” Journal of Francisco H. G. Ferreira, Margaret Grosh, Niall Keleher, Economic Literature 46 (3): 607–68. Pedro Olinto, and Emmanuel Skoufias. World Bank Hungi, Njora, Demus Makuwa, Kenneth Norman Ross, Policy Research Report. Washington, DC: World Bank. Mioko Saito, Stéphanie Dolata, Frank Van Cappelle, Fryer, Roland G., Jr. 2011. “Financial Incentives and Student Laura Paviot, et al. 2010. “SACMEQ III Project Results: Achievement: Evidence from Randomized Trials.” Quar- Pupil Achievement Levels in Reading and Mathematics.” terly Journal of Economics 126 (4): 1755–98. Southern and Eastern Africa Consortium for Monitoring ————. 2017. “Management and Student Achievement: Evi- Educational Quality, Paris. dence from a Randomized Field Experiment.” Working ILO (International Labor Organization). 2015. “Global paper, Harvard University, Cambridge, MA. Employment Trends for Youth 2015: Scaling Up Invest- Garner, Andrew S., Jack P. Shonkoff, Benjamin S. Siegel, ments in Decent Jobs for Youth.” ILO, Geneva. Mary I. Dobbins, Marian F. Earls, Laura McGuinn, John Insel, Thomas R., and Story C. Landis. 2013. “Twenty-Five Pascoe, et al. 2012. “Early Childhood Adversity, Toxic Years of Progress: The View from Nimh and Ninds.” Stress, and the Role of the Pediatrician: Translating Neuron 80 (3): 561–67. Developmental Science into Lifelong Health.” Pediatrics Jacob, Brian A. 2005. “Accountability, Incentives, and Behav- 129 (1): e224-e231. ior: The Impact of High-Stakes Testing in the Chicago Gertler, Paul J., James J. Heckman, Rodrigo Pinto, Arianna Public Schools.” Journal of Public Economics 89 (5): 761–96. Zanolini, Christel Vermeersch, Susan Walker, Susan M. — —— —. 2007. “Test-Based Accountability and Student Chang, et al. 2014. “Labor Market Returns to an Early Achievement: An Investigation of Differential Perfor- Childhood Stimulation Intervention in Jamaica.” Science mance on NAEP and State Assessments.” NBER Working 344 (6187): 998–1001. Paper 12817, National Bureau of Economic Research, Cambridge, MA. Glewwe, Paul W., Eric A. Hanushek, Sarah D. Humpage, and Kiessel, Jessica, and Annie Duflo. 2014. “Cost Effectiveness Renato Ravina. 2011. “School Resources and Educational Report: Teacher Community Assistant Initiative (TCAI).” Outcomes in Developing Countries: A Review of the Lit- IPA Brief (March 26), Innovation for Poverty Action, erature from 1990 to 2010.” NBER Working Paper 17554, New Haven, CT. National Bureau of Economic Research, Cambridge, MA. Kingdon, Geeta. 1996. “The Quality and Efficiency of Private Glewwe, Paul W., Michael Kremer, and Sylvie Moulin. 2009. and Public Education: A Case-Study of Urban India.” “Many Children Left Behind? Textbooks and Test Scores Oxford Bulletin of Economics and Statistics 58 (1): 57–82. in Kenya.” American Economic Journal: Applied Economics Kremer, Michael R. 1995. “Research on Schooling: What We 1 (1): 112–35. Know and What We Don’t, a Comment on Hanushek.” Gove, Amber, and Peter Cvelich. 2011. “Early Reading, Ignit- World Bank Research Observer 10 (2): 247–54. ing Education for All: A Report by the Early Grade Learn- Kremer, Michael R., Edward Miguel, and Rebecca Thornton. ing Community of Practice.” Rev. ed. Research Triangle 2009. “Incentives to Learn.” Review of Economics and Statis- Park, NC: Research Triangle Institute. tics 91 (3): 437–56. Grantham-McGregor, Sally M., Lia C. H. Fernald, Rose M. C. Kuhl, Patricia K. 2010. “Brain Mechanisms in Early Language Kagawa, and Susan Walker. 2014. “Effects of Integrated Acquisition.” Neuron 67 (5): 713–27. Child Development and Nutrition Interventions on Lauwerier, Thibaut, and Abdeljalil Akkari. 2015. “Teachers Child Development and Nutritional Status.” Annals of the and the Quality of Basic Education in Sub-Saharan New York Academy of Sciences 1308 (1): 11–32. Africa.” ERF Working Paper 11, Education Research and Green, Duncan. 2016. How Change Happens. Oxford, U.K.: Foresight, Paris. Oxford University Press. Lavinas, Lena, and Alinne Veiga. 2013. “Brazil’s One Laptop Grindle, Merilee Serrill. 2004. Despite the Odds: The Conten- Per Child Program: Impact Evaluation and Implementa- tious Politics of Education Reform. Princeton, NJ: Princeton tion Assessment.” Cadernos de Pesquisa 43 (149). University Press. Layton, Lyndsey. 2015. “Study: Billions of Dollars in Guilfoyle, Christy. 2006. “NCLB: Is There Life Beyond Test- Annual Teacher Training Is Largely a Waste.” Washington ing?” Educational Leadership 64 (3): 8–13. Post, August 4. https://www.washingtonpost.com/local Hanushek, Eric A. 1986. “The Economics of Schooling: /education/study-billions-of-dollars-in-annual-teacher Production and Efficiency in Public Schools.” Journal of -training-is-largely-a-waste/2015/08/03/c4e1f322-39ff Economic Literature 24 (3): 1141–77. -11e5-9c2d-ed991d848c48_story.html. ————. 1992. “The Trade-Off between Child Quantity and Leder, Gilah C. 1987. “Teacher Student Interaction: A Case Quality.” Journal of Political Economy 100 (1): 84–117. Study.” Educational Studies in Mathematics 18 (3): 255–71. ————. 1995. “Interpreting Recent Research on Schooling in Lemos, Renata, and Daniela Scur. 2016. “Developing Man- Developing Countries.” World Bank Research Observer agement: An Expanded Evaluation Tool for Developing 10 (2): 227–46. Countries.” RISE Working Paper 16/007, Research on Hanushek, Eric A., Victor Lavy, and Kohtaro Hitomi. 2008. Improving Systems of Education, Blavatnik School of “Do Students Care about School Quality? Determinants Government, Oxford University, Oxford, U.K. 32 | World Development Report 2018 Levitt, Steven D., John A. List, Susanne Neckermann, and Chestnut Hill, MA. https://timssandpirls.bc.edu/pirls Sally Sadoff. 2016. “The Behavioralist Goes to School: 2011/international-results-pirls.html. Leveraging Behavioral Economics to Improve Educa- Mullis, I. V. S., M. O. Martin, P. Foy, and M. Hooper. tional Performance.” American Economic Journal: Economic 2016.  “TIMSS 2015 International Results in Mathemat- Policy 8 (4): 183–219. ics.” TIMSS and PIRLS International Study Center, Bos- Levy, Brian, Robert Cameron, Ursula Hoadley, and Vinothan ton College, Chestnut Hill, MA. http://timssandpirls.bc Naidoo. 2016. “The Politics of Governance and Basic Edu- .edu/timss2015/international-results/. cation: A Tale of Two South African Provinces.” Occa- Muralidharan, Karthik, Abhijeet Singh, and Alejandro sional Working Paper 2, Graduate School of Develop- Ganimian. 2016. “Disrupting Education? Experimental ment Policy and Practice, University of Cape Town, Cape Evidence on Technology-Aided Instruction in India.” Town. NBER Working Paper 22923, National Bureau of Eco- Liang, Xiaoyan, Huma Kidwai, and Minxuan Zhang. 2016. nomic Research, Cambridge, MA. How Shanghai Does It: Insights and Lessons from the Muralidharan, Karthik, and Venkatesh Sundararaman. 2011. Highest-Ranking Education System in the World. Directions “Teacher Performance Pay: Experimental Evidence from in Development: Human Development Series. Washing- India.” Journal of Political Economy 119 (1): 39–77. ton, DC: World Bank. Muralidharan, Karthik, and Yendrick Zieleniak. 2013. “Mea- Linden, Leigh L. 2008. “Complement or Substitute? The suring Learning Trajectories in Developing Countries Effect of Technology on Student Achievement in India.” with Longitudinal Data and Item Response Theory.” Edited by Michael Trucano. InfoDev Working Paper 17 Paper presented at Young Lives Conference, Oxford (June), World Bank, Washington, DC. University, Oxford, U.K., July 8–9. Lupien, Sonia J., Suzanne King, Michael J. Meaney, and Neal, Derek. 2013. “The Consequences of Using One Assess- Bruce S. McEwen. 2000. “Child’s Stress Hormone Levels ment System to Pursue Two Objectives.” Journal of Eco- Correlate with Mother’s Socioeconomic Status and nomic Education 44 (4): 339–52. Depressive State.” Biological Psychiatry 48 (10): 976–80. Nelson, Charles A. 2016. “Brain Imaging as a Measure of Mansuri, Ghazala, and Vijayendra Rao. 2013. Localizing Devel- Future Cognitive Outcomes: A Study of Children in Ban- opment: Does Participation Work? Policy Research Report gladesh Exposed to Multiple Levels of Adversity.” Presen- Series. Washington, DC: World Bank. tation, CMU Department of Psychology Colloquium, Martorell, Paco, and Isaac McFarlin Jr. 2011. “Help or Hin- Department of Psychology, College of Humanities and drance? The Effects of College Remediation on Academic Social Sciences, Carnegie Mellon University, Pittsburgh, and Labor Market Outcomes.” Review of Economics and September 29. Statistics 93 (2): 436–54. Nguyen, Trang. 2008. “Information, Role Models, and Per- McCoy, Dana Charles, Evan D. Peet, Majid Ezzati, Goodarz ceived Returns to Education: Experimental Evidence Danaei, Maureen M. Black, Christopher R. Sudfeld, from Madagascar.” Economics Department, Massachu- Wafaie Fawzi, et al. 2016. “Early Childhood Developmental setts Institute of Technology, Cambridge, MA. Status in Low- and Middle-Income Countries: National, OECD (Organisation for Economic Co-operation and Devel- Regional, and Global Prevalence Estimates Using Predic- opment). 2011. Strong Performers and Successful Reformers in tive Modeling.” PLOS Medicine 13 (6): e1002034. Education: Lessons from PISA for the United States. Paris: McEwan, Patrick J. 2015. “Improving Learning in Primary OECD. Schools of Developing Countries: A Meta-Analysis of —— ——. 2016. PISA 2015 Results: Excellence and Equity in Educa- Randomized Experiments.” Review of Educational Research tion. Vol. 1. Paris: OECD. 85 (3): 353–94. Orazem, Peter F. 2000. “The Urban and Rural Fellowship Mingat, Alain, and Jee-Peng Tan. 1998. “The Mechanics of School Experiments in Pakistan: Design, Evaluation, and Progress in Education: Evidence from Cross-Country Sustainability.” Economics of Education Review 22 (3): Data.” Policy Research Working Paper 2015, World Bank, 265–74. Washington, DC. Orazem, Peter F., Paul W. Glewwe, and Harry Patrinos. 2007. Mizala, Alejandra, and Ben Ross Schneider. 2014. “Negotiat- “The Benefits and Costs of Alternative Strategies to ing Education Reform: Teacher Evaluations and Incen- Improve Educational Outcomes.” Department of Eco- tives in Chile (1990–2010).” Governance 27 (1): 87–109. nomics Working Paper 07028, Iowa State University, Morgan, Claire, Anthony Petrosino, and Trevor Fronius. Ames. 2012. “A Systematic Review of the Evidence of the Impact Park, Rufina Kyung Eun. 2016. “Preparing Students for South of Eliminating School User Fees in Low-Income Develop- Korea’s Creative Economy: The Successes and Challenges ing Countries.” Evidence for Policy and Practice Informa- of Educational Reform” [refers to the Republic of Korea]. tion and Co-ordinating Centre, Social Science Research Research Report, Asia Pacific Foundation of Canada, Unit, Institute of Education, University of London. Vancouver. Mulkeen, Aidan G. 2010. Teachers in Anglophone Africa: Issues in Pascale, Richard T., Jerry Sternin, and Monique Sternin. Teacher Supply, Training, and Management. Development 2010. The Power of Positive Deviance: How Unlikely Innovators Practice in Education Series. Washington, DC: World Bank. Solve the World’s Toughest Problems. Boston: Harvard Busi- Mullis, I. V. S., M. O. Martin, P. Foy, and K. T. Drucker. 2012. ness Press. “PIRLS 2011 International Results in Reading.” TIMSS PASEC (Programme d’Analyse des Systèmes Éducatifs de la and PIRLS International Study Center, Boston College, Confemen). 2015. PASEC 2014: Education System Performance Overview | 33 in Francophone Africa, Competencies and Learning Factors in Lopez-Boo, et al. 2015. “Wealth Gradients in Early Child- Primary Education. Dakar, Senegal: PASEC. hood Cognitive Development in Five Latin American Paxson, Christina H., and Norbert R. Schady. 2007. “Cogni- Countries.” Journal of Human Resources 50 (2): 446–63. tive Development among Young Children in Ecuador: Scott-Clayton, Judith, and Olga Rodriguez. 2014. “Develop- The Roles of Wealth, Health, and Parenting.” Journal of ment, Discouragement, or Diversion? New Evidence on Human Resources 42 (1): 49–84. the Effects of College Remediation Policy.” Education Piper, Benjamin, Evelyn Jepkemei, Dunston Kwayumba, and Finance and Policy 10 (1): 4–45. Kennedy Kibukho. 2015. “Kenya’s ICT Policy in Practice: Singh, Abhijeet. 2015. “Learning More with Every Year: The Effectiveness of Tablets and E-readers in Improving School Year Productivity and International Learning Student Outcomes.” FIRE: Forum for International Research Divergence.” CESifo Area Conference on the Economics in Education 2 (1): 3–18. of Education, CESifo Group, Munich, September 11–12. Pradhan, Menno, Daniel Suryadarma, Amanda Beatty, Snilstveit, Birte, Jennifer Stevenson, Radhika Menon, Daniel Maisy Wong, Arya Gaduh, Armida Alisjahbana, and Phillips, Emma Gallagher, Maisie Geleen, Hannah Jobse, Rima Prama Artha. 2014. “Improving Educational Qual- et al. 2016. “The Impact of Education Programmes on ity through Enhancing Community Participation: Learning and School Participation in Low- and Middle- Results from a Randomized Field Experiment in Indone- Income Countries: A Systematic Review Summary sia.” American Economic Journal: Applied Economics 6 (2): Report.” 3ie Systematic Review Summary 7, Interna- 105–26. tional Initiative for Impact Evaluation, London. http:// Pritchett, Lant. 2013. The Rebirth of Education: Schooling Ain’t www.3ieimpact.org/media/filer_public/2016/09/20/srs7 Learning. Washington, DC: Center for Global Develop- -education-report.pdf. ment; Baltimore: Brookings Institution Press. Solano-Flores, Guillermo, Luis Ángel Contreras-Niño, and Pritchett, Lant, and Amanda Beatty. 2015. “Slow Down, You’re Eduardo Backhoff Escudero. 2005. “The Mexican Trans- Going Too Fast: Matching Curricula to Student Skill lation of TIMSS-95: Test Translation Lessons from a Levels.” International Journal of Educational Development 40: Post-mortem Study.” Paper presented at Annual Meeting, 276–88. National Council on Measurement in Education, Mon- Pritchett, Lant, and Justin Sandefur. 2013. “Context Matters treal, April 12–14. for Size: Why External Validity Claims and Development Spaull, Nicholas, and Janeli Kotze. 2015. “Starting Behind and Practice Do Not Mix.” Journal of Globalization and Develop- Staying Behind in South Africa: The Case of Insurmount- ment 4 (2): 161–98. able Learning Deficits in Mathematics.” International Rajkumar, Andrew Sunil, and Vinaya Swaroop. 2008. “Public Journal of Educational Development 41: 13–24. Spending and Outcomes: Does Governance Matter?” Suryadarma, Daniel. 2012. “How Corruption Diminishes the Journal of Development Economics 86 (1): 96–111. Effectiveness of Public Spending on Education in Indo- Rivkin, Steven G., Eric A. Hanushek, and John F. Kain. 2005. nesia.” Bulletin of Indonesian Economic Studies 48 (1): 85–100. “Teachers, Schools, and Academic Achievement.” Econo- Tan, Jee-Peng, and Alain Mingat. 1992. Education in Asia: metrica 73 (2): 417–58. A Comparative Study of Cost and Financing. World Bank Robinson, Viviane M. J., Claire A. Lloyd, and Kenneth J. Regional and Sectoral Studies Series. Washington, DC: Rowe. 2008. “The Impact of Leadership on Student World Bank. Outcomes: An Analysis of the Differential Effects of Tandon, Prateek, and Tsuyoshi Fukao. 2015. Educating the Leadership Types.” Educational Administration Quarterly Next Generation: Improving Teacher Quality in Cambodia. 44 (5): 635–74. Directions in Development: Human Development Rockoff, Jonah E. 2004. “The Impact of Individual Teachers Series. Washington, DC: World Bank. on Student Achievement: Evidence from Panel Data.” Todd, Alexa, and Mark Mason. 2005. “Enhancing Learning in American Economic Review 94 (2): 247–52. South African Schools: Strategies beyond Outcomes- Romer, Paul Michael. 2015. “Botox for Development.” Paul Based Education.” International Journal of Educational Romer’s Blog, September 13. https://paulromer.net/botox Development 25 (3): 221–35. -for-development/. Tooley, James, and Pauline Dixon. 2007. “Private Education RTI International. 2009. “Early Grade Reading Assessment for Low-Income Families: Results from a Global Research Toolkit.” Research Triangle Institute, Research Triangle Project.” In Private Schooling in Less Economically Developed Park, NC. Countries: Asian and African Perspectives, edited by Prachi Sabarwal, Shwetlena, David K. Evans, and Anastasia Mar- Srivastava and Geoffrey Walford, 15–39. Oxford Studies shak. 2014. “The Permanent Input Hypothesis: The Case in Comparative Education Series. Oxford, U.K.: Sympo- of Textbooks and (No) Student Learning in Sierra Leone.” sium Books. Policy Research Working Paper 7021, World Bank, UIS (UNESCO Institute for Statistics). 2006. Teachers and Washington, DC. Educational Quality: Monitoring Global Needs for 2015. Save the Children. 2013. “Ending the Hidden Exclusion: Montreal: UIS. Learning and Equity in Education Post-2015.” Education —— — —. 2016. “Sustainable Development Data Digest: Laying Global Initiative, Save the Children International, the Foundation to Measure Sustainable Development London. Goal 4.” UIS, Montreal. Schady, Norbert R., Jere Behrman, Maria Caridad Araujo, UIS (UNESCO Institute for Statistics) and EFA (Education Rodrigo Azuero, Raquel Bernal, David Bravo, Florencia for All). 2015. “A Growing Number of Children and 34 | World Development Report 2018 Adolescents Are Out of School as Aid Fails to Meet the Us about the Effect of Leadership on Student Achieve- Mark.” Policy Paper 22/Fact Sheet 31, UIS, Montreal; EFA, ment.” McRel Working Paper, McRel International, Paris. Denver. UNESCO (United Nations Educational, Scientific, and Cul- Wolf, Alison. 2004. “Education and Economic Performance: tural Organization). 2013. Third Regional Comparative Simplistic Theories and Their Policy Consequences.” and Explanatory Study (TERCE). UNESCO Regional Oxford Review of Economic Policy 20 (2): 315–33. Bureau for Education in Latin America and the Carib- World Bank. 2003. World Development Report 2004: Making bean, Santiago, Chile. http://www.unesco.org/new/en Services Work for Poor People. Washington, DC: World /santiago/education/education-assessment-llece/third Bank; New York: Oxford University Press. -regional-comparative-and-explanatory-study-terce/. — —— —. 2011. World Development Report 2012: Gender Equality ——— —. 2016. Global Education Monitoring Report 2016, Educa- and Development. Washington, DC: World Bank. tion for People and Planet: Creating Sustainable Futures for All. — —— —. 2013. Service Delivery Indicators (database). World Paris: UNESCO. http://unesdoc.unesco.org/images/0024 Bank, Washington, DC. http://datatopics.worldbank.org /002457/245752e.pdf. /sdi/. USAID (U.S. Agency for International Development). 2017. — —— —. 2014. “STEP Skills Measurement Surveys: Innovative Early Grade Reading Barometer. Washington, DC. http:// Tools for Assessing Skills.” Social Protection and Labor www.earlygradereadingbarometer.org/. Discussion Paper No. 1421. Washington, DC. http:// Uwezo. 2014. “Are Our Children Learning? Literacy and documents.worldbank.org/curated/en/51674146817873 Numeracy across East Africa 2013.” Twaweza, Nairobi. 6065/STEP-skills-measurement-surveys-innovative Wales, Joseph, Ahmed Ali, and Susan Nicolai. 2014. “Improve- -tools-for-assessing-skills. ments in the Quality of Basic Education: Chile’s Experi- — —— —. 2016a. Assessing Basic Education Service Delivery in the ence.” With Francisca Morales and Daniel Contreras. Philippines: The Philippines Public Education Expenditure Case Study Report: Education, Overseas Development Tracking and Quantitative Service Delivery Study. Report Institute, London. AUS6799. Washington, DC: World Bank. Walker, Susan P., Theodore D. Wachs, Julie Meeks Gardner, — —— —. 2016b. “Francophone Africa Results Monitor: Basic Betsy Lozoff, Gail A. Wasserman, Ernesto Pollitt, Julie A. Education (Multiple Countries).” World Bank, Washing- Carter, and the International Child Development Steer- ton, DC. http://documents.worldbank.org/curated/en/doc ing Group. 2007. “Child Development: Risk Factors for search/projects/P156307. Adverse Outcomes in Developing Countries.” Lancet 369 — —— —. 2017. “Driving Performance from the Center: (9556): 145–57. Malaysia’s Experience with Pemandu.” Knowledge and Waters, Tim, Robert J. Marzano, and Brian McNulty. 2003. Research: The Malaysia Development Experience Series, “Balanced Leadership: What 30 Years of Research Tells World Bank, Kuala Lumpur. Overview | 35 Contents of the World Development Report 2018 Foreword Acknowledgments Abbreviations Overview: Learning to realize education’s promise Part I: Education’s promise 1 Schooling, learning, and the promise of education Part II: The learning crisis 2 The great schooling expansion—and those it has left behind Spotlight 1: The biology of learning 3 The many faces of the learning crisis Spotlight 2: Poverty hinders biological development and undermines learning 4 To take learning seriously, start by measuring it Spotlight 3: The multidimensionality of skills Part III: Innovations and evidence for learning Spotlight 4: Learning about learning 5 There is no learning without prepared, motivated learners 6 Teacher skills and motivation both matter (though many education systems act like they don’t) 7 Everything else should strengthen the teacher-learner interaction 8 Build on foundations by linking skills training to jobs Spotlight 5: Technology is changing the world of work: What does that mean for learning? Part IV: Making the system work for learning at scale 9 Education systems are misaligned with learning Spotlight 6: Spending more or spending better—or both? 10 Unhealthy politics drives misalignments 11 How to escape low-learning traps 37 ECO-AUDIT Environmental Benefits Statement The World Bank Group is committed to reducing its environmental footprint. In support of this commitment, we leverage electronic publishing options and print-on- demand technology, which is located in regional hubs worldwide. Together, these initiatives enable print runs to be lowered and shipping distances decreased, resulting in reduced paper consumption, chemical use, greenhouse gas emissions, and waste. We follow the recommended standards for paper use set by the Green Press Initiative. The majority of our books are printed on Forest Stewardship Council (FSC)–certified paper, with nearly all containing 50–100 percent recycled content. The recycled fiber in our book paper is either unbleached or bleached using totally chlorine-free (TCF), processed chlorine–free (PCF), or enhanced elemental chlorine– free (EECF) processes. More information about the Bank’s environmental philosophy can be found at http://www.worldbank.org/corporateresponsibility. Every year, the World Bank’s World Development Report (WDR) features a topic of central importance to global development. The 2018 WDR— LEARNING to Realize Education’s Promise—is the first ever devoted entirely to education. And the time is right: education has long been critical to human welfare, but it is even more so in a time of rapid economic and social change. The best way to equip children and youth for the future is to make their learning the center of all efforts to promote education. The 2018 WDR explores four main themes: First, education’s promise: education is a powerful instrument for eradicating poverty and promoting shared prosperity, but fulfilling its potential requires better policies—both within and outside the education system. Second, the need to shine a light on learning: despite gains in access to education, recent learning assessments reveal that many young people around the world, especially those who are poor or marginalized, are leaving school unequipped with even the foundational skills they need for life. At the same time, internationally comparable learning assessments show that skills in many middle-income countries lag far behind what those countries aspire to. And too often these shortcomings are hidden—so as a first step to tackling this learning crisis, it is essential to shine a light on it by assessing student learning better. Third, how to make schools work for all learners: research on areas such as brain science, pedagogical innovations, and school management has identified interventions that promote learning by ensuring that learners are prepared, teachers are both skilled and motivated, and other inputs support the teacher-learner relationship. Fourth, how to make systems work for learning: achieving learning throughout an education system requires more than just scaling up effective interventions. Countries must also overcome technical and political barriers by deploying salient metrics for mobilizing actors and tracking progress, building coalitions for learning, and taking an adaptive approach to reform. SKU 33155