ENDING LEARNING POVERTY: What will it take? 2 PHOTO BY: © THE WORLD BANK Contents Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 The challenge: End learning poverty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 The vision: Learning for all children and youth. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 The magnitude of the problem—and the crisis at the foundations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 The importance of foundational skills . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Reading—A key foundational skill and a gateway to learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Every child should be reading by age 10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Learning poverty: A new early-warning indicator to spotlight low learning levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Measuring Learning Poverty: Achievements and remaining data gaps ..................................................................................... 13 Ending learning poverty will be hard: Three findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Where we are now: Half of the children in low- and middle-income countries are learning-poor. . . . . . . . . . . . . . . . . . . . . . . 16 Rates by country group ................................................................................................................................................. 16 What we know about differences by gender ...................................................................................................................... 16 Where the world is headed: At current rates of progress, eliminating learning poverty by 2030 is out of reach. . . . . . . . . . 16 An early warning: Even at the fastest rates of progress seen in recent decades, learning poverty will not be eliminated by 2030. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 The way forward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 A new global learning-poverty target to drive action and sharpen our efforts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Interventions focused on literacy can accelerate progress toward the Learning Target and raise overall education quality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Component 1: Ensure political and technical commitment to clear goals, means, and measures for literacy..................................... 22 Component 2: Ensure effective teaching for literacy............................................................................................................ 23 Component 3: Ensure timely access to more and better age- and skill-appropriate texts.............................................................. 25 Component 4: First teach children in the language they speak and understand......................................................................... 27 Adapting the Literacy Policy Package across diverse country conditions .................................................................................. 28 Interventions reflect country capacities and circumstances................................................................................................... 30 Interventions targeted at literacy require a broader agenda for improving education quality . . . . . . . . . . . . . . . . . . . . . . . . . 33 Five pillars of system improvement .................................................................................................................................. 33 Using technology to support literacy efforts....................................................................................................................... 36 Accompanying this country-level support is an ambitious measurement and research agenda. . . . . . . . . . . . . . . . . . . . . . . . 37 A call to action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Annex A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Ending Learning Poverty: What will it take? | 3 Acknowledgments This report was produced by a core team consisting of João Pedro Azevedo, Michael Crawford, Reema Nayar, Halsey Rogers, Maria Rebeca Barron Rodriguez, Elaine Ding, and Marcela Gutierrez Bernal, under the overall guidance of Annette Dixon, Jaime Saavedra, and Omar Arias. The report draws heavily on joint work with Silvia Montoya, Adolfo Imhof, and Friederich Huebler of the UNESCO Institute for Statistics, whom the team thanks for their excellent collabo- ration in building the harmonized data set and for their comments on the analysis. Substantial inputs were provided by Kristoffer Gustav Bjarkefur, Marguerite Clarke, Natasha De Andrade Falcao, Ning Fu, Tihtina Zenebe Gebre, Koen Martijn Geven, Diana Goldemberg, Laura Gregory, Syedah Aroob Iqbal, Maria Jose Vargas Mancera, Sergio Venegas Marin, Harry Anthony Patrinos, Shwetlena Sabarwal, Brian William Stacy, Jason Allen Weaver, Ryoko Tomita Wilcox, and Hongxi Zhao. The team thanks Husein Abdul-Hamid, Dina Abu-Ghaida, Melissa Ann Adelman, Hanna Alasuutari, Rita Almeida, Samer Al-Samarrai, Nina Arnhold, Juan Baron, Roberta Malee Bassett, Tara Beteille, Andreas Blom, Luis Crouch, Amanda Dever- celli, Emanuela Di Gropello, Simeon Djankov, Halil Dundar, Safaa El Tayeb El-Kogali, David Evans, Mourad Ezzine, Amber Gove, Amer Hasan, Robert J. Hawkins, Linda Heibert, Rafael de Hoyos, Victoria Levin, Julia Liberman, Toby Linden, Mes- kerem Mulatu, Ambar Narayan, Elizabeth Ninan, Benjamin Piper, Adelle Pushparatnam, Alfonso Sanchez, Christopher J. Thomas, Michael Trucano, Waly Wane, and Nobuo Yoshida for their inputs and comments. We are also grateful to those whose dedicated work produced all the learning data used in this paper—the national governments and the teams at the IEA (PIRLS and TIMSS), UNESCO (LLECE), Confemen (PASEC), and SACMEQ, as well as the OECD. Finally, we thank the many government counterparts and other partners who have provided feedback and support during the development of this work. We apologize for any omissions and express our sincerest thanks to everyone, whether or not they are named here, who graciously gave their time and expertise. Acknowledgments | Ending Learning Poverty: What will it take? 4 PHOTO BY: © BART VERWEIJ / WORLD BANK Introduction All children should be able to read by age 10. Reading is a gateway for learning as the child progresses through school—and conversely, an inability to read slams that gate shut. Beyond this, when children cannot read, it’s usually a clear indication that school systems aren’t well enough organized to help children learn in other areas such as math, science, and the human- ities either. And although it is possible to learn later in life with enough effort, children who don’t read by age 10—or at the latest, by the end of primary school—usually fail to master reading later in their schooling career. Ending Learning Poverty: What will it take? 5 In recent years, it has become clear that many children it in the foreseeable future requires far more rapid prog- around the world are not learning to read proficiently. ress at scale than we have yet seen. Even though the majority of children are in school, a large proportion are not acquiring fundamental skills. Moreover, To galvanize this progress and strengthen its own 260 million children are not even in school. This is the lead- efforts, the World Bank is: ing edge of a learning crisis that threatens countries’ efforts to build human capital and achievement of the Sustainable 1. Launching a new operational global learning Development Goals (SDGs).1 Without foundational learn- target to cut the Learning Poverty rate by at ing, students often fail to thrive later in school or when they least half before 2030 join the workforce.They don’t acquire the human capital •  Simulations show that this target is ambitious they need to power their careers and economies once they yet achievable if all countries manage to im- leave school, or the skills that will help them become en- prove learning as well as the top performers of gaged citizens and nurture healthy, prosperous families. the 2000–15 period did—which means on av- erage nearly tripling the global rate of progress. As a major contributor to human capital deficits, the learning crisis undermines sustainable growth and pov- 2. Using three key pillars of work to support coun- erty reduction. The Human Capital Project is raising aware- tries to improve the human capital outcomes of ness of the costs of inaction. The average Human Capital their people Index (HCI) score across countries is 0.56; this means that A literacy policy package consisting of inter- •  by the age of 18, a child born today will be only 56 percent ventions focused specifically on promoting ac- as productive as a child would be under the benchmark of quisition of reading proficiency in primary school a complete education and full health.2 Shortcomings in the quality and quantity of schooling, which have been sum-  refreshed education approach to strengthen • A marized as a learning crisis, are a leading contributor to this entire education systems—so that literacy im- human capital deficit. Poor education outcomes have ma- provements can be sustained and scaled up and jor costs for future prosperity, given that human capital is all other education outcomes can be achieved the most important component of wealth globally. Indeed,  n ambitious measurement and research • A its importance grows as countries become more prosper- agenda—covering measurement of both ous: in high-income Organisation for Economic Co-oper- learning outcomes and their drivers and con- ation and Development (OECD) countries, human capital tinued action-oriented research and innovation makes up over 60% of wealth.3 on how to build foundational skills. To spotlight this crisis, we are introducing the concept Change is needed at scale, quickly, and for large of Learning Poverty, drawing on new data developed populations. That cannot be done without technol- in coordination with the UNESCO Institute for Statis- ogy. Open-source digital infrastructure and informa- tics. Learning poverty means being unable to read and tion systems will be used to ensure that resources understand a simple text by age 10. This indicator brings reach all teachers, students and schools. together schooling and learning indicators: it begins with the share of children who haven’t achieved mini- Education initiatives alone are not enough. The fight mum reading proficiency (as measured in schools) and against learning poverty will require an integrated, is adjusted by the proportion of children who are out of multi-sectoral approach supported by actions beyond school (and are assumed not able to read proficiently). the education sector—that is, in all the other areas es- sential to improve learning. For example, ensuring that The new data show that 53% of all children in low- and all children can learn requires better water and sanita- middle-income countries suffer from learning poverty.4 Progress in reducing learning poverty is far too slow to tion, improved health and nutrition, better social protec- meet the SDG aspirations: at the current rate of improve- tion for disadvantaged populations, civil service reforms, ment, in 2030 about 43% of children will still be learn- and strengthened management and financing of pub- ing-poor. Even if countries reduce their learning poverty lic services. All of this requires a whole-of-government at the fastest rates we have seen so far in this century, the approach to better learning outcomes. Beyond this, re- goal of ending it will not be attained by 2030. newed attention is needed to the role that families and communities play in building the demand for education, There is an urgent need for a society-wide commitment creating the right environment for learning, and sup- to invest more and better in people. If children cannot porting the right education reforms. The Human Capital read, it is clear that all education SDGs are at risk. Elimi- Project recognizes this need to work across sectors to nating learning poverty is as important as eliminating ex- bring together all the actions required to improve hu- treme monetary poverty, stunting, or hunger. To achieve man capital. Introduction | Ending Learning Poverty: What will it take? 6 PHOTO BY: © LIANG QIANG / WORLD BANK The challenge: End learning poverty Education is foundational for countries’ growth, productivity, and development; for individ- ual and family incomes and welfare; for improved health outcomes (including reduction in fertility); for active participation in civics and political life; for social cohesion; and for active participation of individuals and societies in the global economy. And in turn, literacy and oth- er basic skills are foundational for all other education outcomes. The new Learning Poverty measure is aimed at spotlighting deficits in literacy and spurring action to ensure that all chil- dren can acquire literacy and other foundational skills. Education is a basic human right, and it is also central to unlocking human capabilities—so it is essential to ensure that the right to education is fulfilled in a meaningful way for all children. Ending Learning Poverty: What will it take? 7 The vision: Learning for The magnitude of the problem— all children and youth and the crisis at the foundations Widespread, high-quality education is now seen The core obstacle to fulfilling this vision is that, despite everywhere as the foundation for development, all the advances in schooling in recent decades, young growth, and poverty reduction. The Sustainable De- adults are leaving school with too little learning and velopment Goals embody very high aspirations for ed- too few of the skills that make them employable, pro- ucation. SDG 4 makes the following commitment: by ductive citizens. This problem has many dimensions, but 2030, the signatories will “Ensure inclusive and equi- the crisis starts at the foundational years of the education table quality education and promote lifelong learning system: opportunities for all.” The very first of the commitments under SDG 4 is Target 4.1, which is to “ensure that all • We are experiencing a global learning crisis.6 In girls and boys complete free, equitable, and quality pri- low- and middle-income countries, various met- mary and secondary education leading to relevant and rics show that roughly half of students are going effective learning outcomes.” In other words, the world through school without acquiring the foundational has committed to achieving universal completion of skills they need. The lack of focus on assuring liter- secondary school for all youth—and with meaningful acy and numeracy in many countries implies that learning—by 2030. The Human Capital Index of the millions of children leave school without these most World Bank also highlights the importance of improv- basic cognitive skills. ing broad-based acquisition of skills for all children by • The rate of improvement in learning is too slow. At the age of 18 as a foundation for subsequent produc- the current rate of improvement, it will be impossi- tivity and growth.5 ble to reach even a target of universal basic literacy and numeracy by 2030—let alone the higher-order This emphasis on education for all children and youth skills that countries aspire to for their children and is well placed. Education is a right and it has great inher- youth. ent value, as well as being a key driver of development, growth, and competitiveness. Our vision at the World • The learning crisis comes on top of continued gaps Bank is therefore a world in which, through education, all in enrollment. Enrollment in early childhood educa- countries prepare their children and youth to succeed as tion is low, and primary completion is not yet uni- citizens in a rapidly changing and uncertain world. This versal. In secondary education, dropout rates are still requires that they: very high, particularly in low-income countries and among girls in some countries—sometimes because • Invest in their people; students do not find value in school, and sometimes because of the financial, social, or cultural barriers • Take action showing that learning really matters to they face. In many low-income countries, demo- them; and graphic pressures are magnifying the challenge of • Commit not only the financial resources, but also keeping school enrollment up. the political and managerial resources necessary to build an education system that serves all. The lack of learning reduces the quality of the labor force in many countries, directly translating into a Providing education that prepares today’s children for shortage of skills. A major component of the Human the future is not a static task. The world is changing at Capital Index is the learning-adjusted years of school a rapid pace, as global trends such as disruptive technol- (LAYS), which combines quantity and quality in a single ogy, climate change, and rapidly evolving demograph- measure that captures the education a child born today ics continuously redefine the skills that learners need to can expect to receive by age 18.7 In many countries, that develop today so they will be productive workers and child will receive less than six learning-adjusted years of engaged citizens tomorrow. Yet some things do not school. As noted above, the LAYS measure accounts for change: to be ready for this challenge, all children need a major share of the gap between the typical country to acquire foundational skills in literacy, numeracy, and and the high-human-capital frontier in the HCI. In every basic reasoning, as well as important socio-emotional region except Sub-Saharan Africa, the quality of educa- skills like conscientiousness and perseverance. These tion contributes more than quantity at this point; even skills are the basis for learning how to learn, and for mak- in Africa, quality and quantity gaps contribute in equal ing the vision a reality. parts.8 Improving learning outcomes is thus crucial for in- creasing human capital and productivity. Building these countries’ human capital requires that their children ac- The challenge: End learning poverty | Ending Learning Poverty: What will it take? 8 quire meaningful skills that will help them stay in school as adults, and higher rates of learning appear to be one of longer and become productive citizens and lifelong the reasons.12 For individuals, simple measures of founda- learners. Without that foundation, many people are leav- tional skills help explain earnings even after controlling ing schools or tertiary institutions without essential cog- for the workers’ years of schooling, in both OECD and nitive, socio-emotional, digital, and technical skills. This emerging economies.13 In low- and middle-income coun- leaves them unprepared for an uncertain world in which tries, better reading ability predicts improved financial the nature of jobs is changing rapidly and adaptability is behaviors, whereas schooling does not.14 Across 48 de- at a premium. veloping countries, the association between female pri- mary schooling and reduced child mortality is two-thirds The Sustainable Development Goals—and specifically higher when schooling leads to more learning.15 Goal 4—show that the international community now recognizes these problems. The new focus on learning Contrary to what is often assumed, rapid change in embodied in the SDGs is a significant advance over the technology or society is likely to make foundational Millennium Development Goals, which promoted in- skills even more important. If schooling has trained an creases in access but not improved learning. individual to do only one specific type of task—even one that seems like a technically advanced task—when auto- But nearly five years into the SDG era, it is time to take mation or globalization makes that task obsolete, that in- stock and make course corrections. Given the depth of dividual can become unemployed or suffer a sharp drop in the learning crisis, there are reasons to question whether earnings. By contrast, someone who has mastered foun- the current targets under SDG 4 are feasible, or whether dational literacy, numeracy, and reasoning ability will be new intermediate targets are needed to spur concrete, more able to adapt and learn new skills. Socio-emotional focused action. To do this, we need to understand why skills like resilience and optimism also matter for adapt- this matters, how far we’ve come, where we’re going, and ability, but overall, workers need cognitive skills. how we can do better. When targets are too easy, they do not provide a real incentive for action; however, when tar- gets are impossible to achieve, they will disappear from Reading—A key foundational policy attention. Targets should be ambitious and should put pressure on all actors of the system, but they have to skill and a gateway to learning be—with a lot of effort—feasible. All foundational skills are important—so why focus on reading? Even the most basic definition of foundational The importance of skills encompasses far more than reading: it also includes foundational skills numeracy, basic reasoning ability, and foundational so- cio-emotional skills, among others. But there are several reasons why we have chosen to focus the Learning Pover- There is a mountain of evidence on the benefits of ed- ty metric and learning target specifically on reading. ucation. For individuals and families, education leads to higher productivity and earnings, poverty reduction, • Reading proficiency is an easily understood metric higher rates of employment, better health outcomes, and of learning. In literate societies around the world, greater civic engagement. For societies, education con- reading has for centuries been at the core of formal tributes to faster innovation and growth, better-function- education. Parents and other stakeholders every- ing institutions, greater intergenerational social mobility, where share an understanding that a school’s first higher levels of social trust, and a lower likelihood of con- task is to ensure that children can read proficiently. flict.9 • Reading is a student’s gateway to learning in every other area. When a child becomes proficient in read- We now are aware that foundational skills such as basic ing, that unlocks the door to the vast knowledge literacy and numeracy are important drivers of these codified in texts of all types. Whether the child takes benefits. Common sense tells us that many of these ben- advantage of that ability will depend on many fac- efits of education stem not from the number of years a tors, including the quality of the school system in lat- student spends sitting in the classroom, but from the er grades, but failure to acquire reading proficiency learning or skills that a student acquires.10 And increas- would clearly hinder the ability to learn throughout ingly, research is substantiating this intuition. The level of that individual’s social and working life. skills in a society predicts economic growth better than the level of schooling does.11 Learning contributes to in- • Reading proficiency can serve as a proxy for foun- tergenerational social mobility in a society too: children dational learning in other subjects, in the same in communities with better schools have higher earnings way that the absence of child stunting is a marker Ending Learning Poverty: What will it take? | The challenge: End learning poverty 9 Figure 1: Correlation of reading scores with math and science scores by assessment and level of aggregation PIRLS-TIMSS (4th Grade) Math PIRLS-TIMSS (4th Grade) Sciencie LLECE (6th Grade) Math LLECE (6th Grade) Science PISA-D (15 Yrs) Science PISA (15 Yrs) Math PISA (15 Yrs) Science Brazil (5th Grade) Math PISA-D (15 Yrs) Math Brazil (9th Grade) Math 0.99 0.97 0.96 0.97 0.95 0.96 0.94 0.95 0.94 0.95 0.93 0.92 0.92 0.91 0.91 0.90 0.88 0.86 00.84 0.82 0.81 0.80 0.72 0.64 0.59 0.57 Student School Country Source: Azevedo and others (2019). Note: For Latin American Laboratory for Assessment of the Quality of Education (LLECE), Program for International Student Assessment (PISA), and PISA for Development (PISA-D), the figure shows the correlation between reading and math/science scores within the given assessment. Correlations between scores on Progress in International Reading Literacy Study (PIRLS) literacy and Trends in International Mathematics and Science Study  (TIMSS) math/science assessments are at the country level. of healthy early childhood development. Systems Every child should be that ensure that all children can read are likely to succeed in helping them learn other subjects as well. reading by age 10 The data bear this out: across countries and schools, proficiency rates in reading are highly correlated Why is learning to read proficiently by age 10 so import- with proficiency in other subjects. For example, the ant as a benchmark? Age 10, when children are expected correlation between a country’s reading score on to be in fourth grade,17 is when many children finish mas- the Progress in International Reading Literacy Study tering “the mechanics” of basic reading in high-performing (PIRLS) assessment and its Trends in International systems. By then, they can decode most words and start to Mathematics and Science Study (TIMSS) math score grow as independent readers.  In many countries, by third is 0.95, and the cross-subject correlations within oth- grade students are “reading to learn” more and have fin- er assessments are also strong (Figure 1). Language ished the intensive phase of “learning to read” that consti- development, which is enhanced by reading skills, tutes “early grade reading.”18 While they still improve their is also nurtured in tandem with the development of reading skills, from this point on it is more through practice a child’s self-regulation, a fundamental socio-emo- in independent reading and less through explicit instruc- tional skill.16 Moreover, as can be seen in Figure 1, tion. Once children have learned to decode and become this correlation is clearly stronger at higher levels of fluent readers, they read faster, and this frees up cognitive aggregations such as schools and countries, which is space for them to focus on text meaning. Faster reading the reporting level of the Learning Poverty indicator. means more practice and very often more enjoyment. In a virtuous circle, more reading improves vocabulary and To sum up, numeracy and other basic skills are vitally background knowledge, improving overall reading skills, important, and nothing here should be taken to suggest which leads to more reading. By contrast, if they do not otherwise—but early reading deserves special attention. obtain good skills as readers by approximately age 10, It is precisely because all those skills are so important that they tend to fall further and further behind, and few catch we need to document and then accelerate the rate of prog- up. Indeed, research from the United States indicates that ress on the most basic of skills: reading. Many countries “without . . . systematic and intensive approach to early in- have already achieved success in teaching reading, and the tervention, the majority of at-risk readers rarely catch up. evidence from these successes can be applied to other con- Failure to read by nine years of age portends a lifetime of texts. Increasing literacy through education system reforms illiteracy for at least 70% of struggling readers.”19 strengthens the capacity of countries to then take on and manage more complex education reforms. If countries get High rates of reading by age 10 go hand-in-hand with that right, the public will be more supportive of reforms, better skills later in life. Theoretically, it could be argued education systems will improve service delivery, and chil- that what matters is skills later in life, and not only at age 10. dren will have the tools to learn in every area of knowledge. But in practice, education systems—and not just individual The challenge: End learning poverty | Ending Learning Poverty: What will it take? 10 Box 1: What constitutes “minimum proficiency” in reading? The UNESCO Institute for Statistics (UIS) is leading the effort to develop internationally comparable indicators and methodological tools to measure progress toward the SDG 4 indicator targets. As part of that process, the Global Alliance to Monitor Learning (GAML) has defined a Minimum Proficiency Level (MPL) for reading at the end of primary, and this MPL serves as the basis for determining shares of students with at least minimum reading proficiency and for comparing levels across assessments and countries. While the definition is still undergoing minor refinements, the core concept is clear from the latest version: “Students independently and fluently read simple, short narrative and expository texts. They locate explicitly-stated information. They interpret and give some explanations about the key ideas in these texts. They provide simple, personal opinions or judgements about the information, events and characters in a text.” In addition to this nutshell statement, intended to be accessible to the nonexpert, the GAML has also proposed a common terminology to describe classifications in the context of the MPL. This is a critical first step toward linking cross-national and national learning assessments with a common benchmark. Working with the UIS, we have used this MPL to build a consolidated global database with 100 countries representing 81% of the population of primary-age children globally. Source: Minimum Proficiency Levels: Described, unpacked and illustrated, accessed at: http://gaml.uis.unesco.org/wp-content/uploads/sites/2/2019/05/ GAML6-REF-2-MLP-recommendations-ACER.pdf learners—that miss this age 10 benchmark do not usually This is why we are launching the Learning Poverty in- catch up later. Nearly all systems that perform well on mea- dicator, which combines shortfalls in school access and sures of adolescent and adult learning (such as PISA, 8th learning in one simple measure. It measures a straight- grade TIMSS, and the PIACC assessment of adult skills) have forward concept: what share of children around the world reading-proficiency rates among their 4th graders that ex- are not able to read a short age-appropriate text with ceed 90%, as do many other high-income countries partici- comprehension around age 10? The reading proficiency pating in PIRLS. For example, the rates in 2016 were 99% in rates used for the learning poverty measure come from Hong Kong (China), 97% in Singapore, 96% in Canada, and an approach that combines recent standardized learning 95% in Germany.20 This is true even of systems that choose assessments (cross-national and national) carried out at not to begin literacy instruction early. Finland famously the end of primary school (see Box 1 for a discussion of does not focus on academic skills in kindergarten, but in- this approach).22 Because these assessments miss out- stead emphasizes play-based learning. And yet by the time of-school children, we discount the calculated proficien- Finnish students are assessed by the PIRLS assessment in cy rate by the share of children who are not enrolled in 4th grade (at the average age of 10.8 years), 98% of them school, thus combining quality and quantity measures have already achieved basic reading proficiency.21 In other of schooling. We count out-of-school primary-age chil- words, Finland—like many other high-income countries— dren as learning-poor for two reasons: (1) empirically, has virtually eliminated learning poverty. they are very unlikely to read proficiently; and (2) from a human rights perspective, the Learning Poverty metric should signal that all children should be both in school Learning poverty: A new and learning to read, and that the absence of either one is a form of poverty. (See Box 2 for a technical definition early-warning indicator to of the indicator.) spotlight low learning levels Like monetary poverty, learning poverty demands ur- gent action. The term “learning poverty” underlines just Better information can help ensure that all children how important achieving at least a minimum proficien- acquire the reading skills they need. If hundreds of mil- cy in reading ability is as a vehicle to a productive, ful- lions of children are not getting foundational reading filling life in the modern world. Just as monetary poverty and other skills when they should, by the end of primary excludes people from economic, social, and political op- school, what can be done about it? First, it is critical to portunity, so too does a lack of basic reading skills. And raise awareness of the problem and build a willingness to the two typically go together: poorer and more disadvan- tackle it. One way to do this is to encapsulate the problem taged children are much more likely to be learning-poor in a simple summary indicator that is easy to understand than their better-off peers. This is morally unacceptable, and track. and it also exacts great economic costs on society. Ending Learning Poverty: What will it take? | The challenge: End learning poverty 11 Box 2: What is Learning Poverty? The Learning Poverty indicator combines the concepts of schooling and learning at the end of primary education, building on indicators of reading proficiency and school enrollment generated in the SDG 4 reporting process. Consider this illustration for a hypothetical country that has gaps in both achievement and attainment: Children aged 10-14 (late-primary) A learning assessment found 25% of pupils 75% of pupils 75% of pupils are proficient are not proficient to be reading proficient But only 20% are out of 80% of children 80% of children are enrolled in primary school (OOS) are enrolled in primary Learning The proficient pupils Poverty represent 60% of all children The non-prof. pupils are All OOS children are considered is a combined measure of 20% of all children non-proficient schooling and learning 60% proficient 40% Learning Poor Learning Poverty is the weighted average of the share of the population below the minimum proficiency level, adjusted by the out-of-school population. LP = [(BMP) x (1-OOS)] +[ 1 x (OOS)] where LP = Learning poverty BMP = Share of children at the end of primary who read at below the minimum proficiency level, as defined by the Global Alliance to Monitor Learning (GAML) in the context of the SDG 4.1.1 monitoring OOS = Out-of-school children, as a share of children of primary school age, and in which all OOS are regarded as being below the minimum proficiency level The learning poverty calculations use data from both cross-national and national large-scale assessments that are judged as being of sufficient quality in terms of design, implementation, comparability, timeliness, frequency, documentation, and access. The goal of “reading by age 10” is an ideal: to achieve it, not only should all children be reading proficiently after three full years in primary education, but they should also have entered school at age 6 or 7. By contrast, our actual measurement of learning poverty is based on cross-national or national assessments that are administered in Grades 4, 5, or 6 and therefore at ages between 10 and about 14. The Learning Poverty results presented here therefore may be a conservative estimate of the extent of the literacy challenge for in-school children, since many children have been tested well after age 10. For most countries, the out-of-school children indicator is built using Adjusted Net Enrollment Rate (ANER) data for primary school from UIS. In a few cases, where those data are inconsistent with other evidence, household surveys are used to estimate the out-of-school indicator. Note: To access the data and code to replicate the Learning Poverty indicator, please visit https://github.com/worldbank/LearningPoverty The challenge: End learning poverty | Ending Learning Poverty: What will it take? 12 Figure 2: Data gaps in measuring Learning Poverty: Share of population of children in countries with no or dated learning assessment data, by region, World Bank lending status, and income level 60 50 40 30 (%) 20 10 0 EAP ECA LAC MNA NAC SAR SSA IDA Blend IBRD Rich Low & All middle-income Region Lending World No data No data + old data Source: Azevedo and others (2019) using the Global Learning Assessment Database (https://github.com/worldbank/GLAD). Note: Low-, middle-, and high-Income data include only assessments since 2010; Old data include assessments from before 2010; Regions: East Asia and Pacific (EAP), Europe and Central Asia (ECA), Latin American and Caribbean (LAC), Middle East and North Africa (MNA), North America (NAC), South Asia (SAR), and Sub-Saharan Africa (SSA); Lending Categories: International Development Association (IDA); International Bank for Reconstruction and Development (IBRD); and IDA/IBRD Blend countries, meaning those that are IDA-eligible based on per-capita income levels but also creditworthy for some IBRD borrowing (Blend). Low- and middle-income includes six high-income IBRD clients. Measuring Learning Poverty: Achievements in high-income countries, 70% of these assessments took and remaining data gaps place in the last four years, but in low- and middle-income countries, the figure is only 35%. Data comparability— Thanks to progress in measuring learning and establish- both within and across countries, as well as over time— ing comparability, the new Learning Poverty indicator also poses a significant challenge. (See Box 3.) These gaps covers four-fifths of the target population. Eighty percent underscore the urgency for action on improving data. of children in low- and middle-income countries live in a country with at least one learning assessment at the end of primary, carried out in the past 8 years, that is of sufficient The challenges are even greater for the assessments quality23 to be used for SDG monitoring.24 We are able to use that are not included in this analysis. Here we focus these assessments to construct a global indicator based on primarily on the learning data used to monitor Learning harmonized proficiency levels only because of the partner- Poverty, which relies on reading assessments in Grades 4, ship with UIS and the efforts of the GAML, described above. 5, and 6. Countries should be and are monitoring learning Such comparisons and global aggregation of learning data across different subjects, as well as in earlier grades (such with population coverage this large were not possible until as Grades 2 and 3) and lower secondary (Grade 9). Many recent years. With the recent improvements, the 80% pop- of the issues discussed here—lack of data coverage, com- ulation coverage rate for learning poverty is much higher parability both within and across countries, and lack of than the coverage of the global monetary poverty indica- coordination—are magnified as the scope broadens. tor when it was first launched. The picture that emerges is of a highly fragmented Yet there remain major gaps in data coverage, and learning assessment system with significant variation these gaps often mean we are flying blind in contexts across regions over time and within countries, in terms where the learning crisis is most acute. While LAC and of coverage, comparability, and frequency. These re- EAP have almost 90 percent coverage, less than half of sults corroborate the findings from UIS (2019), which also children in Sub-Saharan Africa live in a country with a points to similar weaknesses. Going forward, the inter- National Large Scale Learning Assessment (NLSA) or a national community must work together to strengthen cross-national learning assessment of adequate quality to country systems, especially in Sub-Saharan African and be used for this purpose (Figure 2). Differences in cover- in fragile and conflict-affected states. More significant age by income level are also striking: virtually all children innovation on implementation modalities and stronger in high-income countries are in educational systems with coordination among development partners will be criti- such monitoring, while only one-third of those living in cal to support countries in filling this crucial data gap and low-income countries are. Recency of the data also differs: making better use of what is already available. Ending Learning Poverty: What will it take? | The challenge: End learning poverty 13 Box 3: Challenges of data comparability While most rich countries assess their learners at Grade 4, low- and middle-income countries tend to be less consistent and evaluate more of their learners at a later age. (See Figure B3.1.) Moreover, some countries participate in different cross-national assessments; this gives them information on learning outcomes at different grades at each point in time, but it prevents them from using these for comparisons and for tracking progress to inform policies and programs. For example, Chile and Colombia participated in LLECE (which assesses learners in Grade 6) in 2013 and in PIRLS (Grade 4) in 2016, but these assessments cannot be used to provide information on changes in learning between 2013 and 2016. In 2011, Honduras applied the PIRLS assessment, but applied it to Grade 6. This made results noncomparable to those of other PIRLS countries, which applied the assessment in Grade 4. And because the PIRLS questions were based on a different assessment framework, results were also inconsistent with Honduras’ LLECE Grade 6 assessment results from 2006 and 2013. Figure B3.1 Share of students assessed by Figure B3.2 Number of spells with temporally grade level (for countries with assessments) comparable data within cross-national and country income classification assessment programs * 100 99 70 `Total 213 90 80 16 TIMSS 70 80 60 51 23 SACMEQ 25 50 44 39 40 34 20 30 PIRLS 94 20 17 15 10 PASEC 0 1 0 0 0 High-income Low- and middle-income All 11 LLECE** 14 Population of assessed learners by grade (%) 0 50 100 150 200 250 Grade 4 Grade 5 Grade 6 Low- and middle-income All Source: Azevedo and others (2019) using the Global Learning Assessment Database. Note: (*) 17 TIMMS and PIRLS spells have been removed due to lack of comparability over time based on existing documentation. In addition, because PASEC is not comparable over time, although it is comparable across countries within a cycle, we do not use the data from PASEC on seven spells. (**) For LLECE we are using the TERCE-SERCE scale; otherwise those spells would also have to be removed. Comparable data on changes in learning over time are particularly scarce. Even between rounds of a given assessment, there is often a challenge of comparability. Using the cross-national assessment data from the past 20 years, we are able to track 213 episodes of change in learning by the end-of-primary school. But most of these are for high-income countries; for low- and middle-income countries, the number is only 70 episodes (See Figure B3.2.) This limitation reflects the sparsity of the data, as assessment cycles follow five- to seven-year intervals or have been entirely irregular. Moreover, some cross-national assessment programs make significant changes in their scales between rounds or even have design instruments suited only for cross-national comparison within rounds, which results in an inability to monitor progress over time. Ignoring such design details in measurement can make it impossible to use assessment to inform policy. The challenge: End learning poverty | Ending Learning Poverty: What will it take? 14 PHOTO BY: © WORLD BANK Ending learning poverty will be hard: Three findings Ending Learning Poverty: What will it take? 15 Where we are now: Half cy: the learning poverty rate is 80% for IDA and IDA/IBRD blend countries. of the children in low- and middle-income countries What we know about differences by gender are learning-poor Despite the barriers confronting girls in some areas of education, in virtually all countries for which we have The headline number that emerges from this analysis is data, girls have lower rates of learning poverty than that at least 53% of all children in low- and middle-in- boys do. Table 1 shows that girls are, on average 6 per- come countries are not able to read proficiently by age centage points less learning-poor than boys.26 The differ- 10—or even at age 12—when many of them are tested. ence is significantly smaller in Europe and Central Asia This learning poverty rate is much higher than the rate and North America, and largest in the Middle East and of extreme income poverty, which has already been re- North Africa (MNA) and East Asia and the Pacific (EAP). duced to 11%, and is well on the way to a global target (Sex-disaggregated data are not available for South Asia.) of poverty elimination by 2030.25 Yet in the education sphere, one out of every two children in the developing The gender difference is significantly greater in mid- world is not learning to read by late primary school age. dle-income countries and in countries in the middle of And the rate is much higher in some regions: in Sub-Saha- the distribution of learning poverty. While in high-in- ran Africa, learning poverty is close to 87%, or nearly sev- come and low-income countries, differences are quite en times as high as the 13% rate found in the World Bank small, the gap reaches 9 percentage points in lower-mid- client countries in Europe and Central Asia (Figure 3). dle-income countries (Table 1). And gender gaps are sig- nificantly higher for countries with a learning poverty be- There is no reason to accept high rates of learning pov- tween 30% and 70% (Figure 5). erty. Based on the experience of rich countries, it should be possible to reduce the rate to close to zero—just as ab- solute poverty is near zero in those countries. Moreover, there are low- and middle-income countries and regions Where the world is headed: which have more recently made dramatic improvements At current rates of progress, in foundational skills. A generation ago, Vietnam was far eliminating learning poverty from achieving even universal primary schooling, but to- day learning poverty has virtually been eliminated, and by 2030 is out of reach Vietnam’s secondary school students achieve PISA scores at the same level as Germany’s. In the state of Ceará in Bra- How has this learning poverty rate changed in recent zil, the municipality of Sobral reformed the career paths of years? Is it declining rapidly enough to ensure that in teachers and principals and provided basic materials to all, 2030, all children will be proficient in reading by age 10, and within a decade it rose from a rank of 1366 on Brazil’s or at least by the end of primary school? Education Development Index (IDEB), a synthetic indicator of education quality, to a rank of 1 in the country. The answer is an emphatic “No”, given the historical rates of progress. When we look at each of the spells of Rates by country group improvement or decline in individual countries between 2000 and 2018,27 we find that: There are very large differences in Learning Poverty across the developing world. The 53% average for low- • The median annual reduction in learning poverty, and middle-income countries is held down by the levels across all spells, is less than 1 percentage point per in upper-middle-income countries, which average 29% year. With a global learning poverty rate of 53% in learning poverty. But in lower-middle-income countries, 2015, this suggests that unless improvement accel- 55% of children cannot read proficiently, and in low-in- erates dramatically from recent historical patterns, come countries, the rate is 90%. (See Figure 4 for more the world will fall well short of eliminating learning details.) poverty by 2030. Similar patterns are found in the data on non-profi- • In around 20% of the recorded episodes of annu- ciency by World Bank lending status. Even in the IBRD alized change, learning poverty increases. Thus, countries in the database, 40% of children are not read- although global rates of reading have been improv- ing proficiently by late primary. And in the other groups, a ing, there is no guarantee of progress in individual substantial majority of children do not acquire proficien- countries. Ending learning poverty will be hard: Three findings | Ending Learning Poverty: What will it take? 16 Figure 3: Percent of children who are learning-poor in low- and middle-income countries, by region 100 90 87 80 Learning Poverty I(%) 70 63 58 60 53 51 50 40 30 21 20 13 10 0 Low- and EAP ECA LAC MNA SAR SSA middle-income Region Source: Azevedo and others (2019) using the Global Learning Assessment Database (https://github.com/worldbank/GLAD); UIS Enrollment Data; and UN population numbers. Note: The MNA Learning Poverty (as in all other regions) only include low- and middle-income countries. The inclusion of the MNA high-income countries, changes the Learning Poverty in the region to 59%. Figure 4: Percent of children who are learning-poor, by country groups and World Bank lending status 100 90 90 80 80 Learning Poverty (%) 70 60 55 48 50 40 40 29 30 20 9 8 10 0 All High-income Upper-middle- Lower-middle- Lower- Non-borrower IBRD IDA & Blend income income income World Income level Lending type Source: Azevedo and others (2019) using the Global Learning Assessment Database (https://github.com/worldbank/GLAD); UIS Enrollment Data; and UN Population numbers. Table 1: Learning poverty by sex and subgroups, for a subsample of countries Domain Description Male Female High-income 8.4 6.6 Income Upper-middle-income 44.6 39.5   Lower-middle-income 55.1 45.9 Low-income 93.3 93.5 EAP 29.6 21.1 ECA 10.0 8.2 LAC 53.0 48.9 Regions MNA 66.0 56.8   NAC 8.0 7.1 SAR . . SSA 86.4 83.0 World Low- & middle-income 55.5 49.8 World All 43.6 38.9 Source: Azevedo and others (2019) using the Global Learning Assessment Database (https://github.com/worldbank/GLAD); UIS Enrollment Data; and UN population numbers. Note: Gender breakdowns calculated using 91 cross-national learning assessments: LLECE, PASEC, PIRLS, SAQMEC, and TIMSS. All assessment data are from after 2010, except for SAQMEC (for Southern and Eastern Africa), where the data are from the Third Round carried out in 2007. Estimates do not reflect the national learning assess- ments in the data set, because of a lack of sex-disaggregated data. Gender breakdown is not possible for South Asia, due to reliance on national learning assessments that do not systematically report that information. Averages of the male and female columns do not match the global averages reported earlier due to changes in the country composition. Ending Learning Poverty: What will it take? | Ending learning poverty will be hard: Three findings 17 Figure 5: Learning poverty gender gap, by country Burkina Faso South Africa Male Mozambique Female Egypt, Arab Rep. Morocco Argentina Botswana Oman Seychelles Bahrain Ukraine Trinidad and Tobago New Zealand Australia France Belgium Slovenia Macao SAR, China United Kingdom Hong Kong SAR, China Luxembourg Sweden Netherlands 50,2 0 10 20 30 40 50 60 70 80 90 100 Learning Poverty Source: Azevedo and others (2019) using the Global Learning Assessment Database (https://github.com/worldbank/GLAD); UIS Enrollment Data; and UN population numbers. Note: Gender breakdowns calculated using 91 cross-national learning assessments: LLECE, PASEC, PIRLS, SAQMEC, and TIMSS. All assessment data are from after 2010, except for SAQMEC (for Southern and Eastern Africa), where the data are from the Third Round carried out in 2007. • Nevertheless, there have been some cases of rap- countries will still not have reached minimum proficiency id improvement. About 20% of the recorded spells in reading by 2030 (Figure 6).28 show annualized reductions in learning poverty of 2 percentage points or more. Even if this tail of the dis- Progress has been slow because of a lack of commit- tribution reflects some statistical noise, this indicates ment to improve the drivers of learning. As the World that it is possible to make rapid progress (in some Development Report 2018 shows, the classroom ex- cases, through a combination of better learning for perience of too many children around the world is not enrolled students and increased enrollment). We dis- conducive to acquiring literacy or other foundational cuss some of these cases below. skills.29 Young children arrive at school unprepared to learn because of malnutrition and a lack of stimula- Globally, business-as-usual leaves the world far from tion, and sometimes they cannot attend school at all. the goal of eliminating learning poverty by 2030. We Teachers often lack the skills, support, or motivation to can use these estimates to simulate how the popula- teach effectively, and the result is teaching time that tion-weighted learning poverty rate can be expected to is lost or poorly used. Textbooks, learning materials, change between 2015 and 2030. Under a business-as-usu- and technology are missing or poorly integrated into al scenario for the world as a whole—just as for the me- teaching and learning. And school management often dian country—learning poverty falls by less than 1 per- has not been professionalized, leaving principals and centage point per year. Starting from a baseline learning other managers unable or unwilling to address the poverty rate of 53% in 2015, at this rate of progress, about problems in the classroom. These failings in service 43% of late-primary children in low- and middle-income delivery are enabled by a lack of technical capacity in Ending learning poverty will be hard: Three findings | Ending Learning Poverty: What will it take? 18 Figure 6: Learning poverty rate under two scenarios, 2016–30 (simulation) 60.0 Learing Poverty Observed and Simulations (%) 52.7 50.0 43.2 40.0 30.0 26.8 26.4 20.0 10.0 0.0 2015* 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 Business as usual High scenario Halving Note: * 2015 based Global estimate; values from 2016 to 2030 are simulated; Source: Azevedo and others (2019) using the Global Learning Assessment Database; UIS Enroll- ment Data; and UN population numbers. education bureaucracies and a lack of political com- neglect of learning likely constrained the gains that were mitment to make learning for all a priority. And too possible. Now that the international education communi- often, the problems are hidden by a lack of good data ty is focusing more on learning—a shift reflected in Sus- on foundational learning and its immediate causes. tainable Development Goal 4 and the learning indicators being used to track it—progress could well accelerate. An early warning: Even at the How much faster could we reduce learning poverty? To identify what is possible with greater effort, the high sce- fastest rates of progress seen in nario uses the 80th percentile of actual experience. Spe- recent decades, learning poverty cifically, it assumes that every country can reduce learn- will not be eliminated by 2030 ing poverty as quickly as the 80th-percentile country in its region did during the 2000–2015 period, with the better performers continuing to maintain their higher rates of But perhaps the business-as-usual scenario is too pessi- progress.31 As Figure 6 shows, this allows much more rap- mistic. For much of the 2000–2015 period, education sys- id progress. Indeed, it represents—and requires—a near tems as a whole did not focus enough on learning.30 To tripling of the rate of progress, from 0.6 to 1.6 percent- be sure, the substantial improvement in primary school age points per year. Nevertheless, even under this highly completion throughout this period should have reduced optimistic scenario, in 2030 the learning poverty rate will learning poverty: as described above, all children out of still be 27%—considerably higher than the rights-based school count as nonproficient in this statistic, so enroll- target of zero, or even an alternate target set at the sin- ing them should have lowered nonproficiency, as long as gle-digit rates found in many wealthier countries (possi- at least some of them learned to read. But policy makers’ bly 5%). Ending Learning Poverty: What will it take? | Ending learning poverty will be hard: Three findings 19 PHOTO BY: © KHASAR SANDAG / WORLD BANK The way forward We must commit to focusing on education quality and ending learning poverty. Eliminating learning poverty for all children by 2030 would require improvements at a rate and scale that is unprecedented. However, we should continue to strive for this goal. Moreover, eliminating learning poverty while advancing other education goals should be a top development priority, along with the twin goals of ending extreme poverty and promoting sharing prosperity. Ending Learning Poverty: What will it take? 20 This requires a strategy for helping children learn to regions, and if countries above that level maintain their read, guided by a new medium-term operational tar- higher rates of progress. In other words, under this sce- get for eliminating learning poverty, and accompanied nario, every country needs to perform like a country that by system-wide reforms to strengthen and build on all has cut learning poverty rapidly enough to place it among foundational skills. the top 20% of improvers since 2000. At the global level, this implies that the rate of progress in reducing learning poverty will need to be accelerated substantially, nearly A new global learning- tripling from 0.6 to 1.6 percentage points per year. And in some regions, with high levels of learning poverty, coun- poverty target to drive action tries would need to reduce learning poverty by about 2.5 and sharpen our efforts percentage points per year. This is very much a stretch target, and it is also consis- The first step is to set targets to guide and track prog- tent with even higher aspirations. Note that it specifies ress on foundational learning—targets that are fea- reducing “by at least half.” It is crucial for the World Bank sible but that demand more of us. Meaningful action and the countries that it works with to aim for fully elimi- requires the right targets for operational engagement. In nating learning poverty. As the largest external financier this case, given how fundamental the right to education of education, the Bank assumes a responsibility to sup- is, that means good stretch targets—those that demand port countries in reducing learning poverty by at least unprecedented commitment but that, with such a com- half by 2030. But policies are defined and implemented mitment, are attainable. at the national or subnational level, and hence it is polit- ical and financial commitments from countries that will The learning poverty rate is the right type of indicator make it possible to meet that operational target before to use for this purpose. It meets three key criteria for mo- 2030 and move quickly toward ensuring that all children tivating action effectively:32 can read by age 10. • S  implicity: Any stakeholder in education—whether While this is a global target, for concrete action to hap- a teacher, parent, business leader, or finance minis- pen in the classrooms of the world, each country has to ter—can understand what it means to not be able establish its own path, with the objective of eventually to read a simple passage and why slashing learning eliminating learning poverty. Targets can and should be poverty is imperative. At the same time, it is associ- set by countries themselves, but they should be under- ated with other, more comprehensive indicators of pinned by a similar level of ambition. As in the case of the interest, so the simplicity does not compromise va- global poverty target, this global learning poverty target lidity. does not dictate what individual countries should do. The simulations underpinning the target indicate the magni- • R  eplicability: The learning poverty measure is based tude of the changes that are necessary, relative to what on publicly available data and is calculated using a we’ve seen so far in this century. They are based on histori- transparent methodology, making it straightforward cal rates of progress, using the very incomplete data avail- to replicate. able. Each country will need to consider the many factors • M  ovability: With enough effort, it should be possible influencing progress in their country—such as income to reduce learning poverty significantly within just growth, policy shifts, conflict, and migration—and decide a few years—meaning that changes in the indicator what is feasible, and then set its own target and strategy. can be used to gauge the effectiveness of policies Global progress over the next decade will represent the and programs. aggregation of all these country-driven efforts. The bot- tom line is that for the world to be on track toward elimi- The new target: c  ut learning poverty at least by half. nating global learning poverty over the next generation, Based on the analysis described above, a feasible yet am- dramatic improvements are necessary in many countries. bitious target for the World Bank’s operational work with low- and middle-income countries is: Reducing learning poverty is part of a process that strengthens the overall quality of education systems. By 2030, reduce by at least half the share of While the learning poverty target focuses on reducing the 10-year-olds who cannot read. share of children with very low performance in low- and middle-income countries, for most of these countries this This target can be attained if every country matches will drive an improvement in their average performance the rapid improvers. Learning poverty can be cut in half as well. There is strong evidence that to move from a very if all countries achieve progress at the 80th percentile low mean performance to at least a middle level of mean of the post-2000 distribution of gains in their respective performance, countries need to substantially reduce the Ending Learning Poverty: What will it take? | The way forward 21 share of children who are at very low levels of perfor- decode written language. Vocabulary, knowledge of syn- mance. Policies to improve learning among lower-per- tax and grammar, and general background knowledge forming schools and pupils (the tail of the distribution) are fundamental to understanding text meaning; unlike are required to improve learning equitably and to reduce code-cracking skills, however, these are built up contin- unfair inequality.33 uously. Engagement, motivation, and enjoyment drive students to read more, helping students become inde- pendent readers whose knowledge and skills increase as Interventions focused on literacy they read more and more complex texts. can accelerate progress toward Children require explicit instruction to gain these the Learning Target and raise skills and build or sustain motivation; imparting it is the most important task of early grade schooling. The overall education quality World Bank’s Policy Package for helping children learn to read consists of four components that bring focus and The Learning Target is conceptualized as a tool to guide facilitate what countries need to do. They are: (1) ensure and accelerate improvement of literacy and as a cata- political and technical commitment to clear goals, means, lyst for broader improvements in education quality. and measures for literacy; (2) ensure effective teaching The actions that spark improvement for achievement of for literacy; (3) ensure timely access to more and better the target are consistent with and contribute to each of age and skill-appropriate texts; and (4) first teach children the pillars of improved education quality discussed in the in the language they speak and understand. next section. Policy and actions should be grounded in evidence. Pol- Component 1: Ensure political and technical icy actions to improve literacy should be deeply ground- commitment to clear goals, means, and ed in the evidence of how children learn to read. Decades measures for literacy of research have now shed light on this process. Policies The road to success begins with commitment to the and practices that conform to evidence produce results. goal of all children learning to read in primary school. Reading well is a very complex skill that integrates nu- Education systems with high learning poverty are fail- merous subskills, some of which need to be learned be- ing on the fundamental task of securing students’ foun- fore others. Good literacy policies are based on detailed dational skills. Changing the learning trajectory in these knowledge of actions that raise the mastery of specific systems requires a commitment to the goals, means, and subskills and are appropriate in scope and sequence. measures that ensure all students become proficient They pay heed to both specific knowledge reader’s needs readers in the early grades. and to the order in which they are best acquired. They may begin by raising general awareness about words Goals, means, and measures comprise a pathway that through preliteracy skills like rhyming and word play be- guides policy and action. National education authorities fore moving to explicit instruction as formal school be- have the mandate and responsibility to create effective gins. Early grade reading makes children explicitly aware school systems where policies and practices lead to learn- that words are made up of sounds, and that symbols ing for all students. In many cases, a vision of success- (called letters in alphabetic languages) represent those ful learning orients the choice of policies, actions, and sounds. Instruction then shows how letters and sounds milestones that assure the vision becomes a reality. This go together, and helps build vocabulary and the ability sounds obvious, but unfortunately it is not the case in to decode written words. Early grade reading imparts the many systems, where there is no clear pathway to learn- skills that students need to read smoothly, fluently, and ing. Systems that perform well connect the end goal of with comprehension. It seeks to accomplish all this in reading to the means of achieving it and the measures ways that motivate children and promote love of reading that verify success. That is, they define national goals, re- and learning. late interventions to those goals, and measure student and system progress on an ongoing basis. Young learners need specific skills along with positive engagement and motivation. Pathways to proficient Goals for literacy recognize the urgency of having all reading are characterized by continuous interaction of students learn to read so that they can begin to “read mastery of specific tasks, continuous knowledge and to learn” by the end of the early grades. The Learning skill building, and building and maintaining strong mo- Target’s focus on reading for comprehension by age 10 tivation among students. The main specific tasks are is consistent with the culmination of successful literacy sometimes referred to as the “mechanics” of reading; they instruction and learning from the start of school through relate primarily to the “code-cracking skills” needed to about the end of the third grade of primary school. Goals The way forward | Ending Learning Poverty: What will it take? 22 for preprimary will focus on language and print aware- Reading assessment results can mobilize country coali- ness, promoting play-based learning that sets children tions and streamline national education reform. When up to succeed when formal instruction begins in prima- Peru ranked last in the 2012 round of PISA, its students’ ry school. Detailed goals in the first years of primary also poor performance in reading and math made headlines track the sequence of subskills that students need to be- across the country. Reformers in the government used come good readers: knowledge that words are made up this information to mobilize public support for a variety of sounds, letter-sound knowledge, strong recognition of reforms, including investing more in education and and knowledge of basic vocabulary words, and the ability improving teachers’ careers and professional develop- to read with increasing smoothness and understanding. ment.34 It also led to developing measurement capacity Goals become more varied and complex for each succes- that strengthened the overall education system. By 2015, sive grade, culminating in the production of graduates the results of PISA showed substantial improvements in who have the reading skills, vocabularies, and a broad reading comprehension of Peruvian secondary students and deep mastery of the background knowledge they and a reduction in the number of students not reaching need to build their human capital and reach their full po- minimum proficiency.35 tential. Projects to improve early literacy should align with and Assessing performance. Some education systems may build upon national goals and curricula with appropri- not have fully committed to improving literacy because ate scope and sequence. Lao PDR’s Reading Readiness they lack good measures of current performance. Educa- Program (RRP) is a good example. It addresses vocabu- tion authorities may not know how few students are be- lary, print awareness, sounds and phonological aware- coming proficient readers or they may believe that stu- ness, narratives and comprehension, and writing. Each dents are making up lost ground after the early grades. topic is linked to specific learning objectives addressed Or their systems may fail to measure the steps on the over the academic year. The RRP features two teacher-led way to literacy, and therein fail to show where and what shared-reading sessions each week, in which teachers types of actions are needed for improvement. It is essen- embed dialogues with children and help children build tial that assessment systems have well-defined ways to the ability to discuss an interesting word or to retell a inform future instruction based on assessment results. story after it is read. The RRP aligns well with the existing These systems require proper design, implementation, curricula by emphasizing shared-reading routines, yet it documentation, and dissemination of results. Attention provides additional structure to these routines to ensure should also be given to ensuring that the assessment that children’s development of key readiness skills is ex- results are comparable within the country over time, al- plicitly targeted. lowing for consistent measurement of country progress. National goals should be set with an understanding of how students are currently doing, and systems should Component 2: Ensure effective teaching for use the data as a baseline on which to develop achiev- literacy able goals, interventions, and indicators of progress. Teachers in many countries are not providing the types, sequences, and/or amounts of instruction students Measures should start early and point to key summa- need to learn to read. The evidence shows that when tive milestones. “Measuring early” involves two import- students are taught in the right way (content, sequence, ant meanings: early in the policy setting process and early and amount of instruction), nearly all of them learn to in the student’s school career. As mentioned above, pol- read.36 However, many teachers in low-income countries icy makers are best served to formulate policies and ac- lack the skills they need to provide effective instruc- tions from a position of knowledge about current student tion overall, and for reading in particular. This is true for performance. This may be a single measure of reading knowledge of both subject content and pedagogy. For proficiency at the end of the early grades, or more gran- example, in Sub-Saharan Africa, the World Bank’s Service ular information from multiple points on the education Delivery Indicator surveys in six countries show that 84% pathway. Early assessment also needs to be considered of Grade 4 teachers have not reached the minimum level as it pertains to the progress of each individual student. of competence. In Lao PDR, only 2.4% of teachers scored Information on preliteracy subskills such as language and 80% or more on a test of Lao language and math, and the print awareness leads to improvements being made to average score on a test of pedagogy was 52%.37 As indi- impart the skills that prepare students for success in pri- cated, teachers need support to acquire the knowledge mary school. The first “paper-based” test of reading com- and abilities required for effective teaching of reading. prehension—typically administered when children are at or near age 10—should be seen as both a crucial first Successful early grade reading interventions do two indicator of success and the culmination of efforts to have things well: First, they introduce reading materials and children master all the subskills of early reading. teacher guides with step-by-step guidance on what to Ending Learning Poverty: What will it take? | The way forward 23 teach and how to teach. In many settings, books that technology can be a cost-effective alternative to on-site have “tightly structured and effective pedagogy” greatly coaching. In particular, the combination of tablet-based help students learn. These books are in fact lesson plans lesson plans (preloaded with demonstration videos) and for teachers which can be used on a voluntary basis; they e-coaching was as effective as paper-based lesson plans simplify the task of providing instruction by allowing and a reading coach.40 teachers to focus on how to teach rather than on what to teach. They remove the need for teachers to devise plans Behavioral change by teachers in classrooms requires on their own, although teachers should be encouraged to teacher and principal buy-in. Improved lesson plans continuously find creative ways to deliver concepts and need to be delivered by capable, motivated teachers. If ideas. The pedagogies and teacher guides are designed by structured lesson plans are not used in the classroom, or reading pedagogy experts, and they incorporate the find- teachers do not change their behavior after training, per- ings from the science of how children learn to read. They fected technical designs will lead nowhere. Thus, making pay attention to organizing instruction around content, sure that reforms to improve literacy build ownership scope, and sequence of tasks and abilities that children with teachers and principals is critical. To build owner- need to become good readers. Countries such as China ship, clear information on what is expected, appropriate and Vietnam have histories of providing clear guidelines support and accountability for meeting these expecta- for teachers and focused textbooks with clear sequences tions, and interventions to support schools that are strug- of content. This approach is credited with helping these gling should be provided. Teachers become allies if these two countries obtain PISA scores significantly above the changes facilitate their work and if they can observe an predicted scores by income level.38 As education systems impact on the improvement of children’s reading levels.41 improve and teachers increase in preparation and exper- tise, the need for highly structured lessons decreases, and Classroom-focused interventions need systems-level lesson plans become a tool that teachers can build upon support. To sustain these focused actions in the class- according to their knowledge and capacity. room that help all children to read, systems need fair and effective management of the careers of teachers and Second, successful interventions provide a new kind school leaders, as well as clear accountabilities and defi- of “training” or teacher professional development (PD) nitions of roles and responsibilities throughout an educa- that strongly emphasizes practicing specific classroom tion system. skills. This “practice-based” PD supports “lesson fidelity”— the ability of teachers to implement the pedagogical plan Mastery of reading skills requires teaching at the right they are given (e.g., lesson plans with tightly structured level. “ Teaching at the right level” is a phrase that has be- and effective pedagogy)—and offers detailed guidance come shorthand for ensuring that each student is given to aid teachers in low-capacity settings. The World Bank’s the task he or she needs to master in his or her learning Education Policy Approach recommends this kind of progression. It means, for example, that students who are teacher training, highlighting the need to make it con- struggling with letter sounds continue to work on letter tinuous and, when possible, “on-site.”39 Through its new sounds and master them before moving to word reading. initiative, Coach, the World Bank aims to support teach- Teaching at the right level (TaRL) is often associated with ers in the content and delivery of their lessons and train a pedagogical approach that the NGO Pratham has used school leaders to provide on-going coaching. This focus throughout India.  This approach has reached more than of classroom practice training helps teachers apply what 50 million students to date in India, Ghana, and Zambia. they learn in their classrooms immediately. This in turn A set of six randomized evaluations over the last decade permits the consolidation of the practiced skills, as teach- have shown that focusing teaching to the level of the stu- ers observe how students respond to this more focused dent improves learning outcomes in reading and math instruction. Continuous support to teachers in this pro- across a variety of contexts.42  TaRL may have the most cess is essential. Coaching, mentoring, and “communities impact in settings characterized by large classes, poor of practice” models give teachers access to experts and/ teacher capability, frequent teacher and student absen- or senior colleagues who help to keep the focus on im- teeism, or pressures for instruction to move forward be- proving classroom teaching practices. Practical, focused fore students have mastered a given subskill in the learn- professional development and continuous follow-up ing sequence. These pressures may result from mandates through coaching and support are indispensable ele- to cover the entire curriculum in a given school year even ments of improvement. when the curriculum has been shown to be overly ambi- tious for most students and teachers.43 In-class teacher support can be expanded through the use of virtual coaching to provide just-in-time guidance Teachers are the key resource for keeping students on for reading instruction strategies. For instance, a study the learning path, and “teaching at the right level” sig- of a comprehensive intervention in South African public nifies multiple options for assessing and reacting to be schools found that locally designed low-cost integrated sure this happens. These include reading camps for chil- The way forward | Ending Learning Poverty: What will it take? 24 PHOTO BY: © LIANG QIANG / WORLD BANK dren in which volunteers teach during summer or out of riculum, teaching guides (with class scenarios), proposed school periods (India); remedial teaching carried-out by materials for each lesson, and suggested adaptations for government teachers and provision of special learning children with special education needs.45 In Liberia, the materials (India); grouping students by ability instead of successful EGRA Plus program combined structured les- by age (Kenya); and adapting remedial education in one sons for teachers with observation and feedback from or two hours of a regular/contract school day (India).44 literacy coaches. This was part of a general approach that Whether it is through group work, or student-level assess- included curriculum-based measures (CBMs) to carefully ment, home reading, or supplementary classes taught by track student progress. These CBMs include both oral and volunteers, all efforts should point the student to the next written assessments that replace the old pencil and pa- challenge in the learning pathway. per-only tests. At the end of this program, reading scores had increased a very impressive 0.82 standard devia- Adaptive learning software can be used effectively to tions.46 tailor reading materials to the proficiency level of indi- vidual learners. For example, Mindspark centers in India used an adaptive learning software that customized con- Component 3: Ensure timely access to more tent based on the proficiency level and rate of progress and better age- and skill-appropriate texts of each student. When students attended these centers Availability of quality, age-appropriate reading materi- after school for a period of four and a half months, they als is a significant predictor of strong early literacy. Chil- achieved dramatic improvements in reading outcomes, dren who lack access to books or exposure to printed or more than doubling their rate of progress. digital and written materials are more generally at a dis- advantage as they try to learn to read. “Print poverty” has Many countries are adopting a holistic approach to pro- huge consequences on performance: data from the Unit- viding structured lesson plans as a means of improving ed States show that students scoring on the 98th percen- results. In Poland, early literacy textbooks are accompa- tile of tests may read 4.7 million words a year, equivalent nied with a rich teaching framework in the form of a cur- to 67 minutes a day; those scoring in the 10th percentile Ending Learning Poverty: What will it take? | The way forward 25 may only read 51,000 words per year, or 1 minute a day.47 esting, or not aligned with the requirements of effective To achieve fluency, students must be exposed to appro- pedagogy. The Global Book Alliance has raised awareness priate texts and have sustained time to practice reading.48 about the extent and causes of the problem. In Malawi, Partner organizations such as UNESCO and USAID have for example, the Tumbuka and Yao languages each have emphasized the importance of timely access to reading approximately 2.2 million native speakers, yet fewer than materials to improve learning.49 Evidence confirms the 20 book titles are available in either language.56 importance of each child having a book of his or her own during instruction, but in reality students often have to The problem has both short- and long-term dimen- share a book with several other classmates. These books sions. From one perspective, the problem is that existing are especially effective if they are in a language the child books are not getting to classrooms and into the hands speaks at home and best understands (see Component of children. Solutions involve purchasing and distributing 4).50 books, not creating more or better. Getting every child a copy of the main reading textbook for each grade would Books alone are not enough, however. It is the combi- constitute an important step forward. However, building nation of high-quality books, distributed at a 1:1 ratio for strong readers requires more than just the main class- children, and supported by effective teachers and clear room textbook. Children need access to large amounts pedagogical guidelines that has the highest impact.51 of books and texts on various topics at various levels of A randomized control trial of the Kenya Primary Math difficulty. Achieving “one book per student” should be an and Reading Initiative (PRIMR) indicated that the com- interim milestone on the way to making many books per bination of structured lessons, PD, and coaching, plus a child available. book for every child, was more than twice as effective as having only two of these three key inputs.52 In Mongolia, The lack of level-appropriate reading materials for chil- improved provision of books provided in isolation led to dren stems from interrelated problems. One problem is a 0.21 standard deviation improvement in student out- a lack of qualified authors and publishers working in the comes. Teacher training, provided in isolation, did not local language. Another is the insufficient or inappropri- have a statistically significant effect on student outcomes. ate use of book procurement and distribution systems, The combination of the books and teacher training how- which increases the costs of provision. In Guinea, Niger, ever, resulted in impacts greater than the sum of impacts and Chad, over 50% of the books that are printed are lost of the two interventions alone, improving student out- in warehousing, transport, and distribution due to lack of comes by 0.35 standard deviations.53 These results point oversight, accountability, and planning.57 towards a critical policy implication that where inputs are complementary, education investments can deliver a To ensure access to more and better age-appropriate much higher result in combination; higher than the addi- texts that are accessible for all, policies to promote tive effect of interventions applied in isolation. reading must intervene at each level of the book chain.58 This will require: (i) book/title development—with at- Texts and reading practice at home complement what tention to authorship, illustration, cultural relevance of children get at school. Reading skills are not built exclu- book content, and publishing capacities; (ii) access/avail- sively through school-based instruction. While high-qual- ability of books for use by children, including licensing ity formal instruction in primary school is imperative, arrangements that permit wider use, formats that allow home reading and preliteracy activities in early childhood for adaptation, and platforms that share existing titles; education are also essential. The evidence confirms the (iii) coordination of procurement systems involved in the importance of a conducive home environment for liter- purchase of books and textbooks to improve efficiency; acy outcomes. A study from Uganda found that an im- (iv) improved supply and distribution chains to ensure portant factor that influenced learning outcomes in early that texts, once developed, are delivered from the pro- literacy was having reading materials at home.54 A more duction site to the students who are the intended end comprehensive study from the Philippines, Uganda, Mali, users; and (v) effective use of texts for reading instruction and Ethiopia found that the home learning environment and practice both in and out of the classroom. was a predictor of literacy across all contexts, with the most critical component of the home environment being Universal access to textbooks is correlated with strong access to print material.55 literacy in Vietnam. Where both text availability and ear- ly grade reading abilities have both been measured, they In low- and middle-income countries, high-quality, often are closely correlated. In Vietnam, for example, the age-appropriate supplementary reading materials may Young Lives initiative carefully measured a number of be scarce or even absent. A full one-fourth of Malawi’s school-related factors among a cohort of children from students lack access to the teaching and learning mate- lower socio-economic households. Measurements in- rials that support literacy. Even when texts are present, cluded both textbook availability and student progress they may be outdated, not contextually relevant, uninter- in reading. A 2013 Young Lives report found that 97% of The way forward | Ending Learning Poverty: What will it take? 26 students in Vietnam own a Vietnamese textbook. Even only reading competency, but also to provide the foun- among the poorest subset of students59 in the poorest dation to study more complex topics. In fact, home lan- province of Vietnam, 97% reported owning a Vietnamese guage should be used across the curriculum, not just in textbook, and 95% reported owning a math textbook.60 reading. Data from Trends in International Mathematics Vietnam has achieved high levels of enrollment in basic and Science Study (TIMSS) have shown that, with few ex- education in recent years and has undertaken important ceptions, in countries with large proportions of students reforms intended to improve school access, quality, and from homes where the language of instruction is not the equity. Results from PISA 2015 showed that the average language spoken at home, math achievement is lower. Vietnamese student has the same level of achievement as Internationally, 4th graders who had not been taught in the average OECD student.61 their home language had average scores 28 points lower on TIMSS 2011 than 4th graders who had been taught in National campaigns have sparked innovations for im- their home language (477 versus 501).69 proving efficiency in publishing and delivering books. Over the past decade Rwanda has made book provi- To implement home language instruction effectively, sion a priority through teaching and learning materials systems need to train their teachers in students’ home reform. A national campaign organized by the Ministry languages and provide appropriate reading and ped- of Education called “Rwanda Reads” aimed to develop a agogical materials. Countries adopt a variety of policies reading culture in Rwanda. Subsequent initiatives have that incorporate home language instruction in ways tai- mobilized a commitment to literacy and have provided lored to their own needs and priorities. They may differ on essential support to publishers, local authors, and teach- duration of home language instruction and if and when ers. second languages are introduced. In all cases, however, national and local educational planning and budgeting Digital texts and e-readers can complement the use of are needed to effectively incorporate home languages textbooks. The work of the Worldreader NGO in Ghana, into the overall functioning of the education system.70 Kenya, and India has demonstrated that it is possible to Teachers should be trained to teach in the local language make available digital reading materials on phones and and be able to use the materials in a targeted way. Teach- other mobile devices in ways that are both accessible to ing how to read in the first and second language should young readers and cost-effective at scale.62 Literacy apps also be coordinated. Specifically, there is great value in such as Bolo and Feed The Monster utilize game-based leveraging similarities between how languages are writ- mechanisms to increase the engagement of emerging ten when beginning to teach reading in the second lan- readers with reading materials. Further, electronic mate- guage. Ecuador and Mali are two examples of countries rials can be created to include support for learners with that have introduced targeted training programs on bi- special educational needs, for example through larger lingual education.71 New technologies can also facilitate texts, audio, and word tracking features.63 the low-cost development and dissemination of mother tongue titles, offer content tailored to the needs and in- terests of individual learners, and augment the printed Component 4: First teach children in the word with multisensory features, such as audio playback language they speak and understand and word tracking. Children gain reading proficiency if taught in their home language first. Students taught to read in a lan- The Pacific Early Age Readiness and Learning Program guage they do not speak at home have great difficulty (PEARL) is showing significant impact in raising school learning.64 Many become frustrated and disengaged, and readiness and early literacy in Tonga. The program es- they are more likely to leave school early and with less tablished playgroups for children aged 0–5; offered struc- knowledge capital.65 By contrast, research has shown that tured lessons focused on early grade reading; provided students in early grades who are taught in their home training and coaching in the home language for teach- language achieve higher reading comprehension.66 In ers in Grades 1 and 2; and helped governments design, fact, research in Sub-Saharan Africa has indicated that implement, and evaluate interventions. The early grade learning how to read in one’s home language can help reading interventions led to an increase in the percent- students acquire greater skill in their second language in age of students able to read with comprehension by 11 later years: the practice of decoding in a language they percentage points (from 18% reading with comprehen- speak can be applied when they attempt to learn in a sion by Grade 2 to 29%).72 Another successful program second language. 67 Effects appear to persist over a life- that supports home language instruction is the Gambia time, with higher average earnings accruing to students READ project. Children were taught in one of seven native who began their schooling in their home language.68 Us- languages, and teachers were trained and provided with ing the home language to instruct students for the early structured lessons. Additionally, children were regularly years of schooling is an important factor to establish not assessed by teachers and by parents through scorecards. Ending Learning Poverty: What will it take? | The way forward 27 An evaluation of the program found that children under • Institutional capacity. A critical factor for success the home language program read with a higher fluency is how well all parts of an education system work in English in Grades 1–3 compared to the children in oth- together to deliver a high-quality classroom expe- er programs. rience. At the favorable end of the spectrum, qual- ified teachers are routinely present in classrooms Decisions about language-of-instruction policies bal- equipped for learning. They act in ways that show ance important technical and political considerations. they take the mandate for all students to learn seri- A country with many linguistic minorities needs learning ously. They have been selected based on their quali- and teaching materials for each language. Important po- fications, and they have been supported and trained litical and economic considerations can drive the desire by school leaders and pedagogical coaches. In these to have all children be proficient in a single national lan- contexts, technically sound instruction is routinely delivered to all students, and teaching is adjusted to guage. Parents may perceive instruction in a widely spo- meet students’ individual needs, in many cases with ken language such as English or French as key to their the use of technology to support the process. On the children’s economic futures. No single language policy other end of the spectrum, in other systems, teach- fits all circumstances. As policy choices are made, how- ers may lack capacity, they might have been hired ever, it is important to keep in mind the strong evidence on political grounds, absenteeism may be high, and for teaching children to read in their home language. activities may not be properly monitored. Inputs are Good language policies that prioritize home language in- often lacking in these settings; classroom sizes are struction can yield positive outcomes while also helping exceptionally large; and professional development, to preserve cultural traditions and resources. Done well, coaching, and support for teachers are scarce and they can promote better skills in both the home language ineffective. Basic skills for literacy that children can and second language in the long term. learn in a few dozen hours in high-institutional-ca- pacity settings (such as the names of all the letters Niger balanced diversity in languages of instruction and in the given alphabet) may take several times longer considerations of economies of scale. Six national lan- to acquire in low-capacity settings or may not ever guages, plus French, are used as languages of instruction be fully learned. Lack of institutional capacity can in Niger. These six languages were chosen from among all manifest itself in many ways, from policies not being recognized national languages because they had devel- implemented adequately, to actors not aligned for oped orthographies and dictionaries that could be used learning, and to learning resources not being ade- in formal instruction. The six languages used as languag- quately distributed. Resources may be available but es of instruction in Niger covered 97.8% of the popula- inefficiently used. tion. Niger also has a policy for transition from exclusive • Language simplicity and transparency. The difficulty home language instruction to bilingual instruction in the of learning to read in a given country depends on how home language and French. Under these policies, school- many languages are spoken, how much they have in ing gradually shifts from 95% home language instruction common, and how difficult they are to learn. All writ- in the 1st grade to 20% home language instruction by the ing systems represent sounds and ideas with symbols, 6th grade. In the 4th grade, both the home and French but how these map to oral language varies greatly in languages are used in class, each half of the time.73 simplicity and regularity. As a result, children have an easier time learning to read in some languages com- pared to others. Writing systems where symbols map Adapting the Literacy Policy Package across directly and simply to sounds are said to have “shal- diverse country conditions low” or “transparent” orthographies; those with indi- rect mapping and many exceptions are said to have Conditions that favor literacy vary considerably across “deep” orthographies.74 Research on 13 European and within countries. Policies and practices should adapt languages found that learning how to read in English appropriately to these variations. The difficulty of getting (whose “deep” orthography has many exceptions to its all children to read by age 10 depends on several factors. spelling rules) needed on average two or three times Perhaps the most critical factors are the institutional ca- more instruction than learning in languages such as pacity of the national education system, the availability Spanish or Italian. 75 (Spanish and Italian have very of the number of texts and books, and the nature of the regular or “shallow” orthographies: letter-sound corre- language(s) to be learned. Depending on the country’s spondences are often one-to-one, and there are few context along these dimensions, different policies are if any exceptions to learn.) The number of languages implementable and appropriate. As a country develops, spoken nationally matters too, as does the common- the context changes, allowing for different policies to be ality among the languages in both oral and written implemented. forms. The size and nature of the job of making ev- The way forward | Ending Learning Poverty: What will it take? 28 eryone literate therefore will depend on many lan- literacy instruction could pay multiple dividends. As guage-related factors. However, research shows that systems go from low to medium to high institutional while difficult languages take more time to learn, the capacity and their students achieve this foundation- vast majority of children can learn to read if the right al skill, they can start to dedicate more time to other conditions are in place and the right policies are im- subjects (as illustrated in Figure 7). For countries in plemented, regardless of the language being learned. fragile, conflict, and violence-affected (FCV) settings, a first measure would be to gather information on • Text and book availability. Children’s skills as read- literacy and other educational needs of the refugee ers vary with the amount they read, which in turn population (including their language) to then set ap- varies with their opportunities to read. Environ- propriate measures and goals. ments where texts and books are ubiquitous both stimulate reading and provide greater opportunities • Ensuring effective teaching. Detailed guidance for for children to practice the skills they are learning in teachers is a feature of successful education systems school. In Cambodia, where a recent PISA for Devel- in many settings, but embracing direct instruction opment (PISA-D) report showed that roughly 50% of and providing lesson plans is especially important students have to share a textbook in school, perfor- where the literacy task is hard or capacity is low. mance in reading was below the PISA-D average.76 Where conditions are unfavorable and capacity is Text availability extends beyond basic textbook low, policies to develop and have teachers imple- availability. Children growing up where texts are ment clear and simple lessons can anchor the de- scarce have greater difficulty entering a virtuous cir- livery of instruction. When conditions make success cle where initial reading enjoyment sparks motiva- harder, structured lesson plans can make teachers’ tion, and in turn leads to more practice, better skills, jobs easier. In FCV countries such as South Sudan,77 and higher motivation. with a dramatic shortage of teachers, training vol- unteers/community members to act as teachers, Countries sit at given points along these three conti- providing them with highly structured lesson plans nua, each with a unique combination of language, in- or other resources to compensate for their lack of ex- stitutional capacity, and text availability—and literacy perience, and using education technology to com- policy choices need to reflect this. While the overall dif- plement this can be a short-term measure to ensure ficulty of the task varies, experience shows that children that these already disadvantaged children acquire can become literate even in the most challenging condi- foundational skills. Directed instruction, enabled tions if the right policies and practices are implemented. through the use of good structured lesson plans, But this requires being aware of where the country stands focuses on the essentials: the scope and sequence in each area and adapting accordingly. What children are of learning tasks that all students need to become taught, how much instruction they receive, and what proficient. As teachers grow in confidence and ca- form that instruction takes should all be compatible with pacity as early reading instructors, offering greater assessments of difficulties and obstacles. autonomy and choice in the reading curriculum and organization of instruction is warranted. As students The Literacy Policy Package should be adapted to pro- master the basics and become more fluent and capa- mote success in any given context. Policy options tai- ble of independent reading, curricula and teaching lored to country circumstances raise the chances of suc- practices may broaden to promote a greater range cess, no matter where along the continuum countries find and diversity of reading skills and interests. Similarly, themselves. Figure 7 presents this visually. Here are some countries with greater institutional capacity and with examples for each of the four components of the policy a more qualified teaching force have less need for package and how each component could be adapted to structured lesson plans. In such contexts, the lesson a country’s context: plans can turn into supporting tools that indicate the core of the lesson on which teachers can build upon • Committing to clear goals, means, and measures according to their knowledge and capacity. for literacy. Countries need to set clear goals on lit- eracy and map out the measures and means they • Ensuring timely access to more and better age- will use to achieve the goals, based on their current and skill-appropriate texts. Ideally, young readers context. When conditions for literacy are unfavor- build enjoyment of reading by finding a variety of able, the share of total time in school dedicated to texts they like on different subjects of interest. When learning how to read may need to increase. In these this is not possible, teachers can still build student less favorable settings, each “unit of progress” takes enjoyment around mastery of reading skills if each more time to achieve. For example, in contexts such child has a high-quality text to read which suits his as Sub-Saharan Africa where 87% of children cannot or her ability. To achieve reading fluency, students read a simple story, increasing the time dedicated to need to be exposed to appropriate text; in fact, re- Ending Learning Poverty: What will it take? | The way forward 29 Figure 7: Adapting the Literacy Policy Package to different country circumstances Institutional capacity Text availability Low Language simplicity and transparency High Settings less favorable to developing literacy Settings more favorable to developing literacy • More direct "structured instruction" • More variety and choice in lesson Countries Countries • Centralized text production to ensure content for teachers may consider may consider one book per student • More variety of textbooks per student • More classroom time dedicated to • Time more equally divided among all developing reading skills school subjects search suggests that time spent with books is the teach in the local language and be able to use the best predictor of reading success.78 Countries with materials in a targeted way. the least favorable conditions for literacy should fo- cus on the first-order goal of ensuring each child has a text. Creating a single text per grade (which instan- Interventions reflect country capacities and tiates the lessons scope and sequence and teacher circumstances guidance, and matches the level of the average stu- Institutional capacity, especially as it bears on the abili- dent in the class) and focusing energy on ensuring ty to deliver instruction, is a key factor that distinguish- that every student receives a copy of the text should es different context for success in literacy.  Text avail- be the first goal for systems in challenging circum- ability—and to a lesser extent, language simplicity and stances. Even when the task is simplified in this way, transparency—tend to vary along with institutional ca- success can be elusive, but some low-capacity coun- pacity, although not in a strict linear way. Different levels tries are already acting to improve their book supply of institutional capacity, usually correlated with different chain. For instance, through a World Bank-support- levels of learning outcomes, imply different sets of poli- ed project, Cambodia has implemented a “Track and cies that should be implemented. An indicative typology Trace” system to show textbook locations in real time developed for illustrative purposes is presented in Table 2 throughout the ordering and distribution process. to provide more details on how the Literacy Policy Pack- After achieving the basic milestone of one book per age can be adapted. child and improving institutional capacity, systems can move toward supplying classrooms and schools • In settings affected by fragility, conflict, and vio- with large numbers of texts for students to develop lence, schools are likely to be scarce, suffer from reading skills in and out of classrooms. unsafe conditions, and have few or no trained • Teaching children first in the language they speak teachers present. Given the protracted nature of and understand best. Using the home language to conflict, countries like Yemen, the Democratic Re- instruct students for the early years of schooling not public of Congo, and Somalia command our sus- only establishes reading competency; it makes the tained attention and require innovative methods study of more complex topics possible, and it leads for delivering education. Working through imple- to better outcomes for students when they later mentation partners such as UN agencies, commu- read in a second language.79 Countries should adopt nity-based organizations, and nongovernmental policies of home-language instruction that are tai- organizations to deliver interventions can yield a lored to their own circumstances (multiple languag- great impact, as in Yemen and South Sudan. In cir- es spoken and transitioning from home language to cumstances when physical school spaces are lack- national language, among other factors). Issues re- ing, the innovative use of technology, such as virtual lated to home language instruction are acutely felt classrooms, use of mobile phones, and radio broad- by refugee children and by children in fragile, con- casting, should be considered. Volunteer teachers flict, and violence-affected contexts. In these dire can also be deployed, aided by technological train- situations, teachers/volunteers should be trained to ing and delivery tools. Where there are no teachers, The way forward | Ending Learning Poverty: What will it take? 30 education technologies can provide emerging read- average student achievement still has room for im- ers with engaging, educational, game-based appli- provement.   Teacher capacities may be higher, but cations to help learners develop basic literacy skills. classroom practice may be outdated, and schools For example, the XPRIZE for Global Learning and may lack the knowledge or means to introduce more Norad’s EduApp4Syria have demonstrated that it is dynamic teaching and learning materials.  Similarly, possible to deliver effective literacy programs using classrooms may not include the full range of need- new technologies in some of the most challenging ed inputs.  In these settings, interventions can bring educational contexts in Africa and the Middle East.80 greater variety and relevance to reading materials Additionally, accelerated learning programs, such as by strengthening national authorship capacity or the World Bank’s Emergency Basic Education Sup- enhancing home reading.  To improve teaching and port Project in the Central African Republic, can be learning, school leaders can be key in making sure deployed after prolonged periods of conflict. that remediation and “teach at the right level” tech- • For countries with low institutional capacity that niques are known and used.   struggle to deliver high-quality instruction, in- • Countries with high institutional capacity gen- terventions require a focus on providing clear erally have the basics of the four components to guidance to teachers, including structured les- ensure that each child can read by the end of pri- sons where appropriate.   Coaching and skill-based mary school. Their constraints are more nuanced training provide the basis for improving instruction.  and respond to more sophisticated constraints, such A variety of efforts can be undertaken to improve as a lack of screening for disabilities or assessment the number of texts that make it to the classroom.  data not necessarily informing policies and teacher Even when such texts are available, success in get- practices.  For example, in Argentina and Uruguay, ting a book for each child is typically elusive. More- assessment systems are developed, but assessment over, low-capacity countries may have a national lan- results need to inform policies more closely to im- guage in which the bulk of instruction is done that prove school, teacher, and student performance. differs from what most people speak. Policies should Providing a high quality of education to all students be in place to have as many children as possible be- regardless of socioeconomic background is also a gin learning in the language they speak and under- challenge, and countries such as Armenia still strug- stand best. Numerous countries have begun insti- gle to provide quality education to disadvantaged tuting policies whereby children first learn to read in groups, vulnerable populations, and learners with the language they speak and understand best. Some disabilities.82   countries like Mali have made efforts to improve ear- ly grade reading with disappointing results.81  But Measurement and assessment at both the student and others, such as Kenya and Nigeria, are building on system levels are key for all countries. Although the some notable successes (by allocating enough re- type of assessment may vary, what is essential is for as- sources, training teachers, and providing students sessment systems to have well-defined ways to feed in- with appropriate reading and pedagogical material formation on student performance back into the system in their home language) and are now seeking to take to drive decisions. National goals need to be set with an good practices to the national scale. understanding of how students are currently doing, and • Countries with moderate levels of institutional ca- then this data needs to be used as a baseline on which to pacity may be able to provide a textbook for each develop achievable goals and indicators of progress to- child and may not suffer from extreme absentee- ward them. Most school systems where 90% of children ism or drastic teacher shortages.  In countries such learn to read have explicit, concrete, and time-bound as Colombia, a range of conditions are found, but goals for early grade readers.83 Ending Learning Poverty: What will it take? | The way forward 31 Table 2: Suggested interventions for countries with different levels of institutional capacity84 Institutional Potential Menu of Possible Interventions Capacity Constraints Level Ensure political and technical commitment to clear goals, means, and measures for literacy • Utilize mechanisms to collect basic information on literacy and other educational characteristics/ Unsafe schools; needs of refugee populations and host communities (including language) nonformal Ensure effective teaching schools; few or no • Use technology (such as virtual classrooms, mobile phones, radio broadcasting) when there are Fragile, trained teachers; no or too few teachers conflict, and no organized • Deploy volunteer teachers aided by tightly structured and effective pedagogy (e.g. proven lesson violence-af- learning program; plans and detailed teacher guides which ensure that the fundamentals are covered) and techno- fected no teaching and logical tools settings learning materi- • Work with community-based organizations, nongovernmental organizations, etc., to aid in service (e.g., als; language of delivery Yemen) instruction issues • Provide accelerated learning and/or remediation through alternative education programs related to refugee Ensure timely access to more and better (age- and language-appropriate, suitable to the level of the and/or fluid popu- students) texts lations. • Focus attention on access to locally relevant, and quality texts considering through use of technology First teach children in the language they speak and understand • Use community volunteers and technology to teach children in the language they speak at home Ensure political and technical commitment to clear goals, means, and measures for literacy • Promote nationally representative measures of early grade reading skills Absent or not well- Ensure effective teaching trained teachers; • Provide tightly structured and effective pedagogy (e.g. through proven lesson plans and detailed low amount of teacher guides which ensure that the fundamentals are covered) time dedicated to • Provide practical, skills-based, on-site teacher training and focus professional development in Low literacy instruc- structured pedagogy and on delivering instruction institutional tion; books, if • Promote use of basic, in-class techniques to check for understanding and teach to the right level capacity available, are too Ensure timely access to more and better (age- and language-appropriate, suitable to the level of the (e.g., Nige- complex; high students) texts ria) ratios of students • Provide interventions to secure one quality book per child by tackling constraints in the book to books; multiple supply chain local languages; • Emphasis on pedagogical quality of the book, including alignment with national curricula and overly rapid tran- appropriate scope and sequence for early reading sition to national First teach children in the language they speak and understand language. • Implement a clear and evidence-based policy on moving from a home language to the national language Ensure political and technical commitment to clear goals, means, and measure for literacy • Use assessment data to adjust teaching to individual student levels Ensure effective teaching Textbooks are • Provide for increasing autonomy in lesson content and structure more common but • Design coaching and professional actionable training for teachers that allows them to teach supplementary literacy effectively Medium texts may be inad- • Institute an array of remediation techniques tailored to student needs institutional equate; remedia- • Empower and increase school leaders’ abilities to guide professional development to the extent capacity tion unavailable; possible (e.g., Co- students do not Ensure timely access to more and better (age- and language-appropriate, suitable to the level of the lombia) get instruction students) texts at the right level; • Develop authorship capacity and improve the availability of books PD is ineffective; • Provide children and teachers with more quality books to develop reading skills in and out of insufficient class classrooms, when resources allow it time for reading. • Emphasis on pedagogical quality of the book, including alignment with national curricula and appropriate scope and sequence for early reading First teach children in the language they speak and understand • Introduce a language transition policy with supporting pedagogical guidance for teachers Ensure political and technical commitment to clear goals, means, and measures for literacy Assessment data • Integrate early grade reading with overall curricula goals not informing • Develop capacity to participate in international and/or regional assessments instruction; lack Ensure effective teaching of alignment of lit- Higher • Empower and increase school leaders’ ability to guide professional development eracy with overall institutional • Provide more variety and choice in lesson content for teachers (might include complex structured curriculum; inad- capacity lesson plans which provide autonomy) equate support (e.g., Arme- Ensure timely access to more and better (age- and language-appropriate, suitable to the level of the to teachers for nia) students) texts lesson planning; • Provide a greater variety of quality books and materials both in and out of school to incentivize lack of relevant motivation for reading content in reading First teach children in the language they speak and understand material. • Support language transition policy with aligned teacher PD and wider text availability in home languages The way forward | Ending Learning Poverty: What will it take? 32 Figure 8: Combinations of key inputs promote more learning: Evidence from Kenya’s PRIMR Program Causal effect size (standard deviations) 1.40 1.29 Effect size (standard deviations) 1.20 1.11 1.00 0.93 0.80 0.73 0.71 0.56 0.56 0.58 0.60 0.38 0.42 0.40 0.34 0.32 0.30 0.20 0.13 0.17 0.16 0.12 0.03 0.00 Math Kiswahlili English Math Kiswahili English Grade 1 Grade 2 PD and coaching PD, coaching, and books PD, coaching, books, and teachers’ guides Interventions targeted at literacy to success, but when systems can focus on a combination of the four is when literacy improves most. require a broader agenda for improving education quality Five pillars of system improvement For interventions targeted at literacy to operate in a The actions countries take to improve early grade sustained way, broad reforms that ensure the right reading and literacy are consistent with and support elements of the system are in place are needed. For a actions to improve education quality more generally. system to improve on a continual basis requires a merito- Significant overlaps and multiple feedback loops charac- cratic teacher career progression system; basic infrastruc- terize relationships. Clear, explicit, and coherent curricu- ture for learning; well-managed systems that deliver the la, with appropriate guidance for teachers, improve early needed feedback; and inputs for continual improvement. reading and other subjects when used appropriately. The package has to be anchored in a system-wide reform The availability of teaching and learning materials pro- for which countries will need more time and political motes learning outcomes across disciplines, and good commitment. language-of-instruction policies benefit instruction re- gardless of what is being taught. Narrower reforms can The World Bank’s Approach to improving education catalyze broader reforms, or vice versa. Countries with systems is built around five pillars. These five pillars rep- poor track records of implementing complex reforms resent the areas where countries typically have to make may seek to focus more on a defined set of goals for progress in order to craft systems that provide the right foundational skills in the early grades. Other countries experiences to their students. Each pillar reinforces the may have the bandwidth to tackle full systems reform all specific literacy policy package described above, and to- at once. gether they can help achieve the Learning Target. The five pillars focus on learners, teachers, classrooms, schools, In the classroom, all inputs come together to create the and education systems and function as detailed in Figure school experience that determines learning outcomes. 9. Figure 10 shows the alignment of the approach and the Significant evidence shows the complementary and in- literacy policy package. teraction effects that occur when multiple inputs come together in the best way. The strongest example is of • Learners are prepared and motivated to learn: Ef- good curricula taught by capable teachers when students forts to support children’s development and learn- each have a copy of a high-quality textbook in a language ing prior to primary school entry are critical to en- they speak and understand. A cost-effectiveness analysis sure children arrive at school prepared to succeed. on the ingredients of success of Kenya’s PRIMR program Robust evidence from countries of all income levels showed that the option of PD, instructional support, 1:1 confirms that a child’s earliest years are a critical revised books, and teachers’ guides was the most expen- window to intervene and build strong foundations sive, but the entire package together had the most addi- for the future—and that it is especially important tional impact on learning and was the most cost-effective to help children from disadvantaged backgrounds intervention85 (Figure 8). No single component is the key keep up with more advantaged peers during this Ending Learning Poverty: What will it take? | The way forward 33 Figure 9: The education approach’s five pillars Learners are prepared Teachers at all levels Classrooms are Schools are safe and Education systems and motivated to are effective and equipped for learning inclusive spaces are well-managed learn valued • Simple, effective, • Eliminate all types • Career track for • Early Childhood • Meritocratic curriculum of school violence principals Education (ECE) profession • Books and & discrimination in • Clear mandates & • Nutrition • Effective HR supportive schools accountability • Stimulation function of Ministry technology • Students with any • Merit-based • Continous • Coaching & disability receive professional school-based structured the right service bureaucracy professional pedagogy • Minimum level of development infrastructure • Children taught at the right level period. High-quality early childhood education pro- holistic teacher management reforms along the gramming should be a mix of efforts to engage par- five principles mentioned above.86 ents in children’s early learning and enroll children • Classrooms are equipped for learning: Countries in high-quality center-based services that promote need to ensure clear, explicit, and coherent cur- cognitive and socio-emotional skills, help children ricula (aligned with teachers’ abilities to deliv- build language and preliteracy skills, and develop er instruction and students abilities to learn), as the love of learning that can carry them through the well as quality inputs, tools, and interventions to rest of their education. Families, in some cases with translate the curricula into effective learning for the support of targeted policies, play a critical role in all students. Curriculum reform requires a simple providing children with care, stimulation, and nutri- design with clearly determined competencies to be tion inputs for long-term cognitive, emotional, and developed by the students and covered by teachers. physical health. Teachers should follow pedagogies, including re- • Teachers at all levels are effective and valued: medial approaches such as “teach to the right level,” Improving teacher quality rests upon five key that allow them to adapt to the needs of all students, principles. First, make teaching an attractive pro- and they should be able to measure what students fession by improving its status, compensation learn to provide timely feedback and inform policy. policies, and career progression structures. Sec- A robust body of literature exists on the importance ond, ensure preservice and in-service education of basic physical inputs to enable student learn- includes a strong practicum component to ensure ing in school.87 In the case of reading, these inputs teachers are well-equipped to transition and per- of course include reading materials: children need form effectively in the classroom. Third, promote access to reading textbooks and reading materials meritocratic selection of teachers, followed by a at home.88 Digital learning resources can provide probationary period, to improve the quality of the further opportunities both in the classroom and at teaching force. Fourth, provide continuous sup- home. Other enabling conditions include desks, ta- port and motivation, in the form of high-quality bles, chairs, blackboards, and school infrastructure in-service professional development and strong (such as walls, ceilings, roofs, electricity, and connec- school leadership, to allow teachers to improve tivity).89 continually. Fifth, use technology wisely to en- • Schools are safe and inclusive spaces: All children hance the ability of teachers to reach every stu- should be able to learn to read in healthy, safe, and dent, factoring in their areas of strength and de- inclusive learning environments. As the number velopment. Although the Literacy Policy Package of countries falling into crisis, fragility, and violence centers on improving the quality of instruction for increases, so too does the number of children who teachers already in the system, its effectiveness re- attend schools in these contexts—meaning that lies on having well-prepared, motivated teachers ensuring safety is a first priority. Bullying, discrimina- in classrooms. In some countries this might require tion, and violence in schools should be eliminated. The way forward | Ending Learning Poverty: What will it take? 34 Figure 10: Summary of the main areas of intersection of the education approach and the literacy policy package Teachers Classrooms Management 2. Ensure effective teaching for literacy 3. Ensure timely 4. Teach children in 1. Assure political access to more and the language they & technical Continuous in-school Detailed guidance (tightly better age and speak and commitment to practical pedagogical structured & effective skill-appropriate understand literacy support pedagogy) texts and readers Assess learning to inform policy & classroom Teach at the right level intruction If a student doesn’t feel safe, protected, and cared student levels of mastery is essential for the selection of for, it is very difficult for any learning to take place. the next steps in the learning process. This is true at the Families, communities, and teaching staff should be student, class, school, district, national, and even inter- aware of the costs of gender and racial stereotyping, national level. Strong assessment policies operate at all and of stereotyping those with disabilities. Learn- levels. Countries undertake national and internationally ing environments must also be inclusive: teaching comparable benchmarking to know where they are in and learning practices should also support those the aggregate and with respect to other countries, and with reading difficulties or disabilities. In practice, to meet their reporting obligations under the Sustain- this means training teachers and other available re- able Development Goals. Yet the information generated source staff (such as special education teachers or by teachers during a single lesson or at specified sum- teacher assistants supporting inclusive practices) mative milestones is no less critical to the progress of a in skills and knowledge to provide differentiated class or a student. Technical assistance tools such as the teaching and support as needed. World Bank’s Learning and Assessment Platform (LeAP— see next section) emphasize the need to build capacity • Education systems are well-managed: The manage- for all key points along the continuum from the student ment capacity of ministries of education, including level to the system level. in subnational offices and in schools, needs to be able to handle the delivery of a complex service These five pillars must be supported by a strong po- like education to thousands of schools every day. litical commitment to ensuring that all children learn. Management capacity means having the people As the World Development Report 2018 emphasized, with the right skills and motivation working within political challenges can pose an even greater barrier to organizational structures aligned toward supporting learning than the technical ones. The political barriers are learning in the school and in the classroom. Coun- created by the competing interests of key actors affecting tries need clear mandates and accountability, mer- the education system, such as politicians, bureaucrats, it-based selection of personnel, and evidence-based employers, and providers, and to lower those barriers, decision making. Technocracies need pedagogical, countries need committed leadership supported by co- managerial, and leadership capabilities to excel. alitions for learning. These coalitions can help good poli- School principals and leaders also have a part to cies and programs survive changes in administration. play, through setting clear roles and responsibilities for staff, ensuring the meritocratic and transparent High-level political commitment also supports a selection into school positions, and more generally whole-of-government approach to learning. Education providing critical support to the implementation of ministries can’t do it alone: they also need support from components under the literacy policy package. other parts of government and society. For all children to learn well, many other pieces must fall into place. For ex- The area of student assessment shows how key ac- ample, families need resources to send children to school, tions operate at multiple levels. Knowledge of current so jobs and social safety nets are crucial; children need Ending Learning Poverty: What will it take? | The way forward 35 to arrive at school healthy and well-nourished, which ating contexts of user groups working on literacy-related requires well-functioning water, sanitation, health, and initiatives, especially in low-income, low-resource com- nutrition systems; transport systems have to make it pos- munities in developing economies. This will be part of sible for children to get to school in the first place; schools the World Bank’s larger “3D” initiative to support the dis- need electrification to operate effectively; and civil-ser- covery, diffusion, and deployment of new technologies in vice regulations need to support well-functioning edu- education: cation bureaucracies. Effective leadership from the top and political coalitions for learning can ensure that the • Discover: Document and analyze evidence-based and government as a whole views ending learning poverty as promising technologies and the necessary pre-condi- a priority, so that all these agencies can work together. tions to utilize them effectively in education, by main- taining a knowledge base of cost-effective and scal- able technologies and generating impact evidence Using technology to support literacy efforts and guidance on key requirements (such as technol- Efforts to accelerate the rate of progress in reading ogy infrastructure and levels of users’ digital skills proficiency would be incomplete without the consid- required for effective implementation). eration of technology. To reach the Global Learning • Diffuse:  Disseminate knowledge about what works Target, countries will need to disrupt existing models and what doesn’t—and why—in the use of new tech- of reading content development, delivery, and mea- nologies in education by supporting the sharing of surement, and new technologies can be critical to such practical information and know-how with policy efforts. The previous discussion already made reference makers and key stakeholder groups, as well as the de- to some of the ways in which new technologies can be velopment of related communities of practice. This used disruptively to achieve the global learning target: includes pointing at the necessary preconditions through the use of digital texts, readers, and didactic and enabling environments for their adoption, giv- materials, as well as apps and other adaptive software en the resource and political economy constraints of applications to teach students to read; computer-fa- schools and education systems today. cilitated student assessment; open digital education platforms and information systems to expand access to • Deploy: S upport the implementation of technolo- reading material; “Track and Trace” technologies to pro- gy-enabled educational products and services, includ- vide real-time monitoring of textbook distribution; and ing by tackling market and procurement barriers virtual coaching for the at-scale delivery of teacher pro- for adoption of the required supporting technolo- fessional development. Underpinning all of these is the gy infrastructure (hardware, connectivity, software, establishment of foundational open-source digital infra- electricity), ensuring effective integration with the structure that will enable the delivery of at-scale solu- curriculum and classroom instruction (e.g., teach- tions and accelerate the rate of progress. As an example, ers’ and students’ digital skills), and using innovative India’s National Digital Infrastructure for Teachers, DIK- tools and approaches to support quick learning and SHA, aims to enable, accelerate and amplify solutions iteration in technology-enabled educational initia- to advance teacher professional development and sup- tives. port. The Bank is supporting similar efforts in countries like Costa Rica and Peru in partnership with the EkStep In partnership with DFID and the Gates Foundation, Foundation. the World Bank is building an ambitious global “Ed- Tech Hub” to implement this 3D strategy. Through this To support all of these initiatives, the World Bank will partnership, the World Bank will help countries adopt help countries find technological solutions that build technologies that support teachers in implementing on effective teaching and bolster literacy in developing “Teach at the Right Level” practices, facilitate the deploy- countries. Interventions that incorporate a smart use of ment of self-paced learning tools for those environments new technologies can have some of the biggest impacts where there is scarcity of effective teachers, and enable on learning. World Bank procurement and advisory work management capacity to deploy these solutions at scale. on procurement activities will help countries avoid the The new EdTech Hub seeks to galvanize a global com- ineffective and expensive adoption of new technolo- munity toward impact, focusing on providing evidence gies in education by generating and diffusing evidence to fund and pursue what works—and to avoid what does of cost-effective uses of technology, as well as shaping not. It will focus on “discovery” and “diffusion” under the the investment decisions of suppliers of literacy-related 3Ds, seeking to complement ongoing World Bank work products and services to meet specific needs and oper- on “deployment.” The way forward | Ending Learning Poverty: What will it take? 36 Accompanying this comparable data for this indicator. This collaboration has included work on developing a global reporting scale country-level support is an and supporting protocols that allow countries to report ambitious measurement “minimum proficiency” data from a variety of national and cross-national assessments in a comparable way. and research agenda This harmonization effort is allowing for a much clearer picture of the learning crisis to emerge. While countries will lead the action to reduce learning The World Bank will also provide client countries with poverty, with country-level support from the World technical support and financing to design and imple- Bank, it is also important to continue to provide mea- ment quality learning assessments. Countries might de- surement and action-oriented research that can sup- cide to generate their own high-quality national assess- port these efforts. Because they benefit many countries, ments or participate in cross-national assessments. The some investments—notably, in the development of mea- World Bank will support the design and implementation surement tools and research—can most cost-effectively of high-quality, fit-for-purpose learning assessments that be carried out at the global level. can also generate internationally comparable data on learning. It will also support the local capacity building First, to reach the Learning Target, countries need to required for continuous assessment and monitoring. Fi- start by knowing where they stand in terms of early lit- nally, knowledge products and capacity-building tools eracy and other foundational skills such as numeracy. will be developed as a global public good. For instance, It is almost impossible for countries to design effective the Bank is working with UIS to develop an internation- early literacy policies without knowing the magnitude al item bank which can be used by countries to comple- and characteristics of the learning poverty in their coun- ment their learning assessments, improve their quality, tries. Furthermore, data can help in building coalitions of and make them internationally comparable. stakeholders (parents, teachers, principals) for improving literacy, understanding what is or is not working, and But the data needs go well beyond just improving stu- tracking how fast progress is being made. dent learning data in primary school. For one thing, they have to start earlier, in the years before a child en- This means improving student learning assessment sys- ters school. Evidence from a range of disciplines confirms tems. Not all countries gather data on learning, and even that a child’s earliest years are a critical time to invest to when data are collected, there are often serious chal- build human capital. The returns to investments in the lenges with their quality. Also, most data from national early years are diverse, and policies and programs to assessments is not internationally comparable. While an improve children’s development have been steadily ex- increasing number of countries participate in cross-na- panding in most countries. Yet despite improvements tional assessments, this does not necessarily contribute in access, in many countries, the quality of early-years to improved national assessment capacity. And some programs is below what is required to promote child de- countries that do measure learning systematically do so velopment. Investments in quality early childhood edu- only at the end of secondary school, when it is often too cation can help children arrive at primary schools with late to make a difference. In fact, over the past decade, strong foundations to succeed and are critical to tackle only about 100 countries have collected internationally learning poverty. The World Bank is working with gov- comparable data for foundational literacy. Having good ernments to scale up these investments; in response to assessment data that allow a country to know if it is on government demand, our investments in early childhood track to meet its goals is essential. education have more than doubled in the last five years. Governments, development partners, and parents also To leapfrog in learning measurement, the World Bank need better information on children’s development and is launching the Learning Assessment Platform (LeAP). the quality of early learning environments. The World LeAP, the activities of which are currently funded by the Bank’s efforts to scale up measurement in early child- second Russia Education Aid for Development (READ) hood include working with partners to scale up global Trust Fund program, aims to improve the quality and monitoring and leading innovative efforts to support a availability of global learning data by aligning cross-na- better measurement of child outcomes and the quality tional learning assessments and expanding national ca- of early learning settings. We are working with govern- pacity to assess learning. As part of LeAP, the Bank estab- ments to integrate early childhood measurement into lished a partnership with UNESCO Institute for Statistics their systems and to use these data to prioritize strategic (UIS), the institution in charge of monitoring SDG 4.1. investments in early learning. Through this partnership, the Bank is collaborating close- ly with UIS and its Global Alliance to Monitor Learning Strengthening adult literacy and other skills is also to ensure that countries are able to report high-quality, essential. Rates of learning poverty were even higher Ending Learning Poverty: What will it take? | The way forward 37 for past generations of children than for today’s, leav- ing up rapidly afterward. At the same time, the GEPD is ing many adults unable to read proficiently (or to read enabling the piloting of new areas of measurement, such at all). According to UNESCO, 750 million adults in the as the prototyping of an EdTech Readiness Index that world have difficulty reading and/or writing. The me- could be used to inform countries of where they stand ga-trends of rapid technological change, demographic on EdTech. To do this, it will measure the extent to which transformation, and global integration of production education technologies, as well as the efforts of multiple imply that good jobs increasingly require novel mix- actors within a larger EdTech ecosystem, are integrated es of cognitive, technical, and socio-emotional skills. with broader education system policies and practices The World Bank’s Skills Toward Employment and Pro- (such as teacher training, curriculum, and infrastructure), ductivity (STEP) measurement program has been pio- and therefore the extent to which investments in EdTech neering efforts to measure and analyze the impact of are likely to bear fruit. these different skills on the socioeconomic success of working-age adults. The World Bank is also reviewing Measurement efforts need to use a “whole child devel- the science of adult literacy acquisition and an analysis opment” lens. Given the evidence on the importance of of adult literacy programs (ALPs) around the world. To socio-emotional skills in education policies, indicators on complement these efforts on adult skills measurement, students’ socio-emotional skills need to be collected in a the WBG and UNESCO Institute for Statistics (UIS) will culturally robust manner. The Bank is working with part- use their recently signed partnership to develop the ners to develop instruments assessing socio-emotional protocols for integrating the adapted version of the skills that, once developed, would freely be made avail- Literacy Assessment and Monitoring (mini-LAMP) into able for use by policy makers, researchers, and organiza- regular household surveys. This could be a significant tions interested in generating performance metrics of the milestone for measuring the skills of adult populations education system with a “whole child” approach lens. This in low-income countries. work will also feed into the World Bank’s existing work on measuring and improving teaching practices around de- To measure the drivers of learning, the World Bank has veloping students’ socio-emotional skills in the classroom launched the Global Education Policy Dashboard (GEP- (through the Teach tool). D).90 This tool will enable countries to monitor how well their practices (or service delivery), policies, and politics The World Bank will also continue to support an ac- are oriented toward learning and attainment for all chil- tion-oriented agenda in research and innovation re- dren. It will measure the quality of key school-level ingre- lated to foundational skills. Among other topics, this dients of learning (teaching, school management, inputs research agenda will explore the knowledge and imple- and infrastructure, and prepared learners), as well as the mentation gaps in policy interventions to improve liter- deeper systemic drivers in policies and politics. This tool acy in middle- and low-income countries (e.g., What are brings together streamlined versions of existing measure- effective models for developing teachers’ content and ment tools such as Teach, a classroom observation tool pedagogical knowledge? What are the best ways to en- for measuring the quality of teaching practices. The GEPD sure that books reach children? How are interventions will generate and report information on a comprehen- best adapted for contexts with different home languag- sive, and yet focused, set of indicators to offer countries es? What adaptations work for fragile settings?). It will also some guidance on where to act and the ability to monitor explore the cognitive processes required for children to progress in the short and medium term. The GEPD indi- read with comprehension and how they can inform policy cators and instruments have been the result of extensive making in client countries (with different languages and collaboration across the education, health, social protec- contexts). Impact evaluations of scalable programs, rapid tion, governance, and other global practices of the World assessments of existing evidence, and lessons from pro- Bank, and the GEPD is being implemented in 13 countries gram implementation processes will be used to provide across all regions in 2019 and 2020, with the goal of scal- client countries with timely and evidence-based advice. The way forward | Ending Learning Poverty: What will it take? 38 PHOTO BY: © SIMONE D. MCCOURTIE / WORLD BANK A call to action The high rates of learning poverty and slow rate of progress in eliminating it are morally and economically unacceptable. All children should learn to read by age 10, both to ensure that they have opportunities in life and to strengthen skills in their societies. Learning poverty places children’s future and the development of their countries at risk. Eliminating learning poverty is an urgent development objective, one that is critical to achieving our goals of end- ing extreme poverty and advancing shared prosperity, as well as any other education goals. We are at an inflection point: the 10-years-olds of 2030 will be born next year. It is critical to accelerate efforts now to ensure that these children will be able to read when they turn 10 years of age. Ending Learning Poverty: What will it take? 39 Tackling this crisis will require a new level of commit- whole-of-government approach, and society-wide ment, coupled with comprehensive reforms to ensure commitment to investing in people. Interventions fo- domestic resources are used as effectively as possible. cused on literacy can accelerate progress towards the Education is a sector with critical macro implications: learning target, but for them to operate in a sustained education spending amounts to 15 to 20 percent of the way, the package has to be anchored in system-wide public budget in many countries, and the education sec- reform. The commitment to ending learning poverty tor builds the human capital that drives development. should have support from ministries of finance, plan- The World Bank is establishing a Global Education Fi- ning ministries, and others with economy-wide respon- nance Platform to support country efforts to ensure that sibilities. Further, for education outcomes to improve, education systems are adequately funded and resources we need interventions from all sectors. Nutrition and are used effectively. The Platform comprises a set of ini- sanitation interventions improve children’s health and tiatives and activities to develop tools, build the evidence foster brain development; cash transfer programs have base, and provide technical assistance. a proven impact on children and particularly girls’ school attendance; transport interventions can reduce costs of A first step is to increase efforts to support countries attending schools; infrastructure investment is needed in measuring learning and implementing systemic ed- to close the gap on the supply side; and digital develop- ucation reforms. Some countries that have data do use ment strategies support school connectivity, enabling it to its fullest to inform policy makers. Unfortunately, EdTech interventions such as adaptive learning. Parents, many countries are flying blind and have very scattered, communities, potential employers, and civil society all inconsistent data. And in several countries, especially in have a role to play. Sub-Saharan Africa, learning is not measured at all.91 Meeting the Learning Poverty Target won’t be easy, but Reforms to end Learning Poverty require interven- we can’t back down from the challenge. We owe it to tions focused on improving literacy, systematic and the children of this world to set our sights high, so they sustained improvements in education systems, a can too. A call to action | Ending Learning Poverty: What will it take? 40 Annex A Table A: Learning poverty by country Country Name Out-of-School Below Minimun Learning Poverty Assessment Year Assessment (OoS) Proficiency (in School) Afghanistan 49.6 87.0 93.4 2013 NLA Argentina 0.6 53.6 53.9 2013 LLECE Armenia 7.2 30.0 35.0 2015 TIMSS Australia 3.2 5.5 8.6 2016 PIRLS Austria 0.0 2.4 2.4 2016 PIRLS Azerbaijan 5.0 19.2 23.3 2016 PIRLS Bahrain 2.1 30.6 32.1 2016 PIRLS Bangladesh 4.9 56.0 58.1 2017 NLA Belgium 1.3 5.1 6.4 2016 PIRLS Benin 3.6 77.3 78.2 2014 PASEC Botswana 7.2 44.3 48.3 2011 PIRLS Brazil 2.7 46.9 48.4 2013 LLECE Bulgaria 6.8 5.2 11.7 2016 PIRLS Burkina Faso 31.7 78.6 85.4 2014 PASEC Burundi 2.7 92.7 92.9 2014 PASEC Cambodia 2.6 49.8 51.1 2013 NLA Cameroon 5.2 75.9 77.2 2014 PASEC Canada 0.0 4.3 4.3 2016 PIRLS Chad 21.1 97.0 97.7 2014 PASEC Chile 9.3 30.3 36.8 2013 LLECE China 0.0 18.2 18.2 2016 NLA Colombia 6.9 44.7 48.6 2013 LLECE Congo, Dem Rep 63.2 62.0 86.0 2011 NLA Congo, Rep 12.8 82.9 85.1 2014 PASEC Costa Rica 1.1 31.7 32.5 2013 LLECE Cote d’Ivoire 21.1 77.6 82.3 2014 PASEC Croatia 3.0 1.0 4.0 2011 PIRLS Cyprus 2.2 14.3 16.2 2015 TIMSS Ending Learning Poverty: What will it take? | Annex A 41 Country Name Out-of-School Below Minimun Learning Poverty Assessment Year Assessment (OoS) Proficiency (in School) Czech Republic 0.0 3.0 3.0 2016 PIRLS Denmark 1.0 2.6 3.6 2016 PIRLS Dominican Republic 6.6 79.4 80.7 2013 LLECE Ecuador 1.9 62.1 62.8 2013 LLECE Egypt, Arab Rep 1.4 69.2 69.6 2016 PIRLS Ethiopia 14.0 88.7 90.3 2015 NLA Finland 0.9 1.7 2.6 2016 PIRLS France 0.9 6.3 7.1 2016 PIRLS Georgia 0.4 13.5 13.8 2016 PIRLS Germany 0.2 5.5 5.7 2016 PIRLS Guatemala 10.1 63.6 67.3 2013 LLECE Honduras 17.1 69.4 74.7 2013 LLECE Hong Kong SAR, China 1.9 1.4 3.2 2016 PIRLS Hungary 3.1 2.9 5.9 2016 PIRLS India 2.3 53.7 54.8 2017 NLA Indonesia 2.4 33.8 35.4 2011 PIRLS Iran, Islamic Rep 0.9 35.1 35.7 2016 PIRLS Ireland 0.0 2.3 2.3 2016 PIRLS Israel 2.9 9.0 11.7 2016 PIRLS Italy 1.4 2.1 3.5 2016 PIRLS Japan 1.2 1.0 2.2 2015 TIMSS Jordan 4.0 50.0 52.0 2015 TIMSS Kazakhstan 0.3 1.9 2.2 2016 PIRLS Korea, Rep 2.7 0.3 3.0 2015 TIMSS Kuwait 3.3 49.4 51.0 2016 PIRLS Kyrgyz Republic 1.9 63.8 64.5 2014 NLA Latvia 3.2 0.8 4.0 2016 PIRLS Lithuania 0.3 2.7 3.0 2016 PIRLS Macao SAR, China 1.3 2.4 3.7 2016 PIRLS Madagascar 21.9 95.8 96.7 2015 NLA Malaysia 1.4 11.7 12.9 2017 NLA Mali 33.0 86.6 91.0 2012 NLA Malta 2.4 26.8 28.6 2016 PIRLS Mexico 1.2 42.5 43.2 2013 LLECE Morocco 5.4 63.8 65.8 2016 PIRLS Netherlands 0.3 1.3 1.6 2016 PIRLS New Zealand 1.5 10.0 11.4 2016 PIRLS Nicaragua 1.6 69.3 69.8 2013 LLECE Annex A | Ending Learning Poverty: What will it take? 42 Country Name Out-of-School Below Minimun Learning Poverty Assessment Year Assessment (OoS) Proficiency (in School) Niger 38.9 97.9 98.7 2014 PASEC Norway 0.2 5.8 6.0 2016 PIRLS Oman 1.5 40.9 41.8 2016 PIRLS Pakistan 27.3 65.0 74.5 2014 NLA Panama 7.1 64.1 66.6 2013 LLECE Paraguay 10.8 71.3 74.4 2013 LLECE Peru 4.2 53.7 55.7 2013 LLECE Poland 4.4 2.0 6.3 2016 PIRLS Portugal 3.6 3.0 6.5 2016 PIRLS Qatar 2.2 33.8 35.3 2016 PIRLS Romania 6.9 14.1 20.0 2011 PIRLS Russian Federation 2.4 0.9 3.3 2016 PIRLS Saudi Arabia 2.5 36.7 38.3 2016 PIRLS Senegal 25.7 65.2 74.1 2014 PASEC Serbia 0.8 7.4 8.1 2015 TIMSS Singapore 0.1 2.7 2.8 2016 PIRLS Slovak Republic 2.1 6.6 8.5 2016 PIRLS Slovenia 2.2 3.7 5.8 2016 PIRLS South Africa 8.4 77.9 79.8 2016 PIRLS Spain 1.5 3.4 4.9 2016 PIRLS Sri Lanka 0.9 14.0 14.8 2015 NLA Sweden 0.4 1.9 2.3 2016 PIRLS Thailand 2.0 21.9 23.5 2011 TIMSS Togo 8.5 84.2 85.6 2014 PASEC Trinidad and Tobago 1.3 19.7 20.7 2016 PIRLS Tunisia 0.4 65.1 65.3 2011 TIMSS Turkey 5.0 17.6 21.7 2015 TIMSS Uganda 9.0 81.1 82.8 2014 NLA United Arab Emirates 2.8 32.4 34.3 2016 PIRLS United Kingdom 0.2 3.2 3.4 2016 PIRLS United States 4.1 3.9 7.9 2016 PIRLS Uruguay 0.5 41.4 41.7 2013 LLECE Vietnam 0.6 1.1 1.7 2011 NLA Yemen, Rep 18.9 93.5 94.7 2011 TIMSS Ending Learning Poverty: What will it take? | Annex A 43 References Anderson, R. C., Paul Wilson, and Linda Fielding. 1988. “Growth Crouch, Luis, and Caine Rolleston. 2017. “Raising the Floor on in Reading and How Children Spend Their Time Outside Learning Levels: Equitable Improvement Starts with the of School.” University of Illinois at Urbana-Champaign. Tail.” RISE Insights. Angrist, Noam, Simeon Djankov, Pinelopi Goldberg, and Harry Crouch, Luis, and Martin Gustafsson. 2018. “Worldwide Patrinos. 2019. “Measuring Human Capital.” World Bank Inequality and Poverty in Cognitive Results: Cross- Policy Research Working Paper No. 8742. sectional Evidence and Time-based Trends.” RISE Working Azevedo, Joao Pedro, Viviane Sanfelice, and Minh C. Nguyen. Paper, January 2018. 2012. “Shapley Decomposition by Components of a Duc, Le Thuc, and Tran Ngo Minh Tam. 2013. “Why Children Welfare Aggregate.” MPRA Working Paper 85584. in Vietnam Drop out of School and What They Do After.” Azevedo, Joao Pedro, and others. 2019. “Will Every Child Be Working Paper 102. Oxford: Young Lives. Able to Read by 2030? Why Eliminating Learning Poverty European Commission. 2012. “Final Report: EU High Level Will Be Harder Than You Think, and What to Do About It.” Group of Experts on Literacy.” Luxembourg. World Bank Working Paper. Washington, DC: World Bank. Filmer, Deon, Amer Hasan, and Lant Pritchett. 2006. “A Banerjee, Abhijit, Rukmini Banerji, James Berry, Esther Duflo, Millennium Learning Goal: Measuring Real Progress in Harini Kannan, Shobhini Mukerji, Marc Shotland, Education.” Center for Global Development Working and Michael Walton. 2017. “From Proof of Concept to Paper 97. Scalable Policies: Challenges and Solutions, with an Filmer, Deon, Halsey Rogers, Noam Angrist, and Shwetlana Application.” Journal of Economic Perspectives, 31 (4): Sabarwal. 2018. “Learning-Adjusted Years of Schooling 73–102. doi: 10.1257/jep.31.4.73 (LAYS): Defining A New Macro Measure of Education.” Beteille, Tara, and David Evans. 2019. “Successful Teachers, World Bank Policy Research Working Paper 8591. Successful Students: Recruiting and Supporting Society’s Washington, DC.: World Bank Group. Most Crucial Profession.” Accessed at: http://documents. Friedlander, Elliott. 2013. “Environmental Factors Associated worldbank.org/curated/en/235831548858735497/ with Early Reading Achievement in the Developing World: Successful-Teachers-Successful-Students-Recruiting-and- A Cross-National Study”. International Journal of Education Supporting-Society-s-Most-Crucial-Profession.pdf Research 55: 25–38. Bulat, J., Dubeck, M., Green, P., Harden, K., Henny, C., Mattos, Glewwe, Paul, Eric Hanushek, Sarah Humpage, and Renato M., Pflepsen, A., Robledo, A., and Sitabkhan, Y. 2017. What Ravina. 2011. “School Resources and Educational Works in Early Grade Literacy Instruction (Knowledge Outcomes in Developing Countries: A Review of and Practice in International Development, 1). RTI Press Literature from 1990 to 2010.” NBER Working Papers Publication No. OP-0039-1702. Research Triangle Park, 17554, National Bureau of Economic Research, Inc. NC: RTI Press. https://doi.org/10.3768/rtipress.2017. Google and Bolo. 2019. “Initial Pilot Findings: March 2019.” op.0039.1702 Infographic. https://bolo.withgoogle.com/static/assets/ Castles, Anne, Kathleen Rastle, and Kate Nation. 2018. “Ending the Bolo_Initial_Pilot_Findings.pdf Reading Wars: Reading Acquisition From Novice to Expert.” Gove, Amber, and Anna Wetterberg. 2011. The Early Grade Psychological Science in the Public Interest 19(1), 5–51. Reading Assessment. RTI Press Book Series. Research Chetty, Raj, Nathaniel Hendren, and Lawrence F. Katz. 2016. Triangle Park, NC. “The Effects of Exposure to Better Neighborhoods on Gove, Amber, and Peter Cvelich. 2011. “Early Reading: Igniting Children: New Evidence from the Moving to Opportunity Education for All. A Report by the Early Grade Learning Experiment.” American Economic Review 106 (4): 855–902. Community of Practice.” Revised Edition. Research Chetty, Raj, Nathaniel Hendren, Patrick Kline, and Emmanuel Triangle Park, NC: Research Triangle Institute Saez. 2014. “Where Is the Land of Opportunity? The Graham, Jimmy, and Sean Kelly. 2018. “How Effective Are Early Geography of Intergenerational Mobility in the United Grade Reading Interventions? A Review of the Evidence.” States.” Quarterly Journal of Economics 129 (4): 1553–1623. World Bank Policy Research Working Paper 8292. Crabbe, Richard, Mary Nyingi, and Helen Abadzi. 2014. Washington, DC.: World Bank Group. “Textbook Development in Low Income Countries.” World Guadalupe, César, Juan León, José Rodriguez, and Silvana Bank Working Paper. Vargas. 2017. “Estado de la Educación en el Perú. Análisis Crawford, Michael, and others. 2019. “Attaining the Learning y perspectivas de la educación básica.” Grupo de Análisis Target: Policies to Promote Literacy for All Children.” para el Desarrollo (GRADE) and Proyecto Fortalecimiento Washington, DC: World Bank Working Paper. de la Gestión de la Educación en el Perú (FORGE).” Lima. References | Ending Learning Poverty: What will it take? 44 Hanushek, Eric A., Guido Schwerdt, Simon Wiederhold, and Patrinos, Harry, and Eduardo Velez. 2009. “Costs and Benefits Ludger Woessmann. 2015. “Returns to Skills around the of Bilingual Education in Guatemala: A Partial Analysis.” World: Evidence from PIAAC.” European Economic Review International Journal of Educational Development. Pages 73: 103–30. 594–596. Hanushek, Eric, and Ludger Woessmann. 2012. “Do Better Piper, Benjamin. 2010. “Uganda Early Grade Reading Schools Lead to More Growth? Cognitive Skills, Economic Assessment Findings Report: Literacy Acquisition and Outcomes, and Causation,” Journal of Economic Growth Mother Tongue.” Research Triangle Park, NC: Research 17(4), pages 267–321. Triangle Institute. J-PAL. 2019. “Teaching at the Right Level to improve learning: Piper, Benjamin. 2018. “Understanding Whether and How Reorienting instruction has improved learning the Tusome Program Worked: Evidence from the opportunities for over 60 million students in India National Scale-up of a Tested Literacy Program in Kenya.” and Africa.” Case Study. Accessed at https://www. Presentation made at CIES 2018. Retrieved on 12 July povertyactionlab.org/case-study/teaching-right-level- 2019. improve-learning Piper, Benjamin, Stephanie S. Zuilkowski, and Salome Ong’ele. Kaffenberger, Michelle, and Lant Pritchett. 2017. “More School 2016. “Implementing Mother Tongue Instruction in the or More Learning? Evidence from Learning Profiles from Real World: Results from a Medium-Scale Randomized the Financial Inclusion Insights Data.” RISE Working Paper Controlled Trial in Kenya.” Comparative Education Review (May). 60 (4): 776–807. Kotze, Janeli, Brahm Fleisch, and Stephen Taylor. 2019. Piper, Benjamin, Stephanie Simmons Zuilkowski, Margaret “Alternative Forms of Early Grade Instructional Coaching: Dubeck, Evelyn Jepkemei, and Simon King. 2018. Emerging Evidence from Field Experiments in South “Identifying the Essential Ingredients to Literacy Africa.” International Journal of Educational Development and Numeracy Improvement: Teacher Professional 66: 203–213. Development and Coaching, Student Textbooks, and Koval-Saifi, Nedjma, and Jan Plass. 2018. “Feed the Monster: Structured Teachers’ Guides.” World Development 106, pp. Impact and Technical Evaluation.” World Vision Inc. and 324–336. Foundation for Information Technology Education and Pritchett, Lant. 2013. The Rebirth of Education: Schooling Development, Inc. Ain’t Learning. Washington, DC: Center for Global Kraay, Aart. 2018. “Methodology for a World Bank Human Development; Baltimore: Brookings Institution Press. Capital Index.” World Bank Policy Research Working Paper Results for Development. 2016. “Final Report: Feasibility Study 8593. for a Global Book Fund.” Kucirkova, Natalia, Catherine Snow, Vibeke Grøver, and Rolleston, Caine, Zoe James, and Le Thuc Duc. 2013. “Making Catherine McBride-Chang. 2017. The Routledge Progress: Report of the Young Lives School Survey in International Handbook of Early Literacy Education: A Vietnam.” Young Lives Working Paper. Oxford Department Contemporary Guide to Literacy Teaching and Interventions of International Development, University of Oxford. in a Global Context. Saavedra, Jaime. 2019a. “Measuring Learning to Avoid Flying Lange, Glenn-Marie, Quentin Wodon, and Kevin Carey. Blind.” World Bank blog published on Education for Global 2018. The Changing Wealth of Nations 2018: Building Development. January 10. Retrieved on September 3, a Sustainable Future. Washington, DC.: World Bank 2019. Group. Saavedra, Jaime. 2019b. “The Most Important Political Decision: Lyon, G. Reid, and Vinita Chhabra. 2004. “The Science of Keep Politics out of the Classroom. Or Not?” World Bank Reading Research.” Educational Leadership 61(6): 12–17. blog published on Education for Global Development. Macdonald, Kevin, Sally Anne Brinkman, Wendy Jarvie, Myrna May 29. Retrieved on October 4, 2019. Machuca-Sierra, Kristen Mcdonnal, Souhila Messaoud- Seymour, Philip, Mikko Aro, and Jane Erskine. 2003. Galusi, Siosiana Tapueluelu, and Bin Thanh Vu. 2017. “Foundation Literacy Acquisition in European “Pedagogy versus School Readiness: The Impact of Orthographies.” British Journal of Psychology, 94, 143–174. a Randomized Reading Instruction Intervention and Shaywitz, Sally. 2003. Overcoming Dyslexia: A New and Complete Community-Based Playgroup Intervention on Early Grade Science-based Program for Reading Problems at Any Level. Reading Outcomes in Tonga.” World Bank Policy Research New York, NY, US: Alfred A. Knopf. Working Paper No. 7944. Washington, DC.: World Bank Shin, Jaran, Misty Sailors, Nicola McClung, P. David Pearson, Group. James V. Hoffman, and Margaret Chilimanjira. 2015. “The Martin, Michael, Ina Mullis, Pierre Foy, and Martin Hooper. 2017. Case of Chichewa and English in Malawi: The Impact of PIRLS 2016 International Results in Reading.  First Language Reading and Writing on Learning English Maurer, Bruno. 2010. “Les langues de scolarisation en Afrique as a Second Language.” Bilingual Research Journal 38 (3): francophone : Enjeux et repères pour l’action. Rapport 255–74. général.” Agence Universitaire de la Francophonie. Paris, Skibbe, Lori, Janelle Montroy, Ryan Bowles, and Frederick France. Morrison. 2019. “Self-Regulation and the Development MoEYS. 2018. “Education in Cambodia: Findings from of Literacy and Language Achievement from Preschool Cambodia’s Experience in PISA for Development.” Phnom through Second Grade.” Early Childhood Research Penh. Quarterly 46 (1st Quarter): 240–251. Oye, Mari, Lant Pritchett, and Justin Sandefur. 2016. “Girls’ Taylor, Stephen, and Marisa von Fintel. 2016. “Estimating the Schooling is Good, Girls’ Schooling with Learning Is Impact of Language of Instruction in South African Better.” Background Paper for The Learning Generation. Primary Schools: A Fixed Effects Approach.” Economics of The Education Commission. Education Review 50: 75–89. Ending Learning Poverty: What will it take? | References 45 The Education Commission. 2016. Achieving a Learning USAID. 2016. Landscape Report on Early Grade Literacy. Generation. Washington DC: United States Agency for International Trends in International Mathematics and Science Study Development. (TIMSS). 2011. TIMSS 2011 International Results in Valerio, Alexandria, María Laura Sánchez Puerta, Namrata Raman Mathematics Chapter 4: Home Environment Support for Tognatta, and Sebastián Monroy-Taborda. 2016. “Are Mathematics Achievement. There Skills Payoffs in Low- and Middle-Income Countries? UNESCO Institute for Statistics. 2017. “More Than One-Half of Empirical Evidence Using STEP Data.” Policy Research Children and Adolescents Are Not Learning Worldwide.” Working Paper 7879, World Bank, Washington, DC. Fact Sheet No. 46 (September). World Bank. 2018a. World Development Report 2018: Learning to UNESCO Institute for Statistics. 2018a. Global Education Realize Education’s Promise. Washington, DC: World Bank. Database. World Bank. 2018b. Growing Smarter: Learning and Equitable UNESCO Institute for Statistics. 2018b. “Global Initiative on Out Development in East Asia and Pacific. Washington, DC: of School Children. South Sudan Country Study.” World Bank. UNESCO Institute for Statistics. 2019. “Improving the World Bank. 2018c. Facing Forward, Schooling for Learning in Monitoring of Learning Outcomes for SDG 4.1.1.” Africa. Washington, DC: World Bank. USAID. 2015. “Planning for Language Use in Education: Best World Bank. 2018d. The Human Capital Project. Washington, DC: Practices and Practical Steps for Improving Learning.” World Bank. EdData II Brief. Washington DC: United States Agency for World Bank. 2018e. “Education Improvement Project in International Development. Armenia.” References | Ending Learning Poverty: What will it take? 46 Endnotes 1 World Bank 2018, UNESCO Institute for Statistics 2017. 2 Kraay 2018, World Bank 2018d. 3 Lange, Wodon, and Carey 2018. 4 This paper draws heavily on two background papers: Azevedo and others 2019, Crawford and others 2019. Note that “Low- and middle-income coun- tries” consists of those defined as IDA, Blend, and IBRD countries under the World Bank lending classification. Note that it includes six IBRD countries that now qualify as high-income: Chile, Croatia, Panama, Poland, Trinidad and Tobago, and Uruguay. 5 Angrist, Djankov, Goldberg, and Patrinos 2019. 6 World Bank 2018a. 7 Filmer, Rogers, Angrist, and Sabarwal 2018. 8 Note that this is true even though there is a lot of room for quantity improvements: The quantity benchmark for a full education in the HCI is 14 years of schooling, meaning that any 18-year-old who has not stayed in school from pre-primary (2 years of pre-primary, in fact) through upper-secondary gradu- ation has not achieved the benchmark. 9 For a recent summary, see World Bank 2018a, Chapter 1. 10 Pritchett 2013. 11 Hanushek and Woessmann 2012, World Bank 2018a. 12 Chetty and others 2016, Chetty and others 2014. 13 Hanushek and others 2015, Valerio and others 2016. 14 Kaffenberger and Pritchett 2017. 15 Oye, Pritchett, and Sandefur 2016. 16 Skibbe and others 2019. 17 In virtually every country, the official start age for primary school is at or before 7 years of age (World Bank Open Data). 18 PIRLS also assesses the reading achievement of young students in their fourth year of schooling as this is “an important transition point in their develop- ment as readers. Typically, by this time in their schooling, students have learned how to read and are now reading to learn” (IEA 2016). 19 Lyon and Chhabra 2004, citing Shaywitz 2003. 20 Mullis, Martin, Foy, and Hooper 2017. 21 As explained below, minimum proficiency on PIRLS is measured as scoring at least 400 points, which PIRLS defines as the Low International Benchmark. 22 Only assessments for Grades 4, 5 and 6 were included in this analysis. 23 Quality is assessed in this context in terms of design, implementation, comparability, frequency, timenliness, documentation, and data access. 24 If we count the number of countries with adequate learning assessments for the learning poverty indicator, rather than using this population-weighted figure, then coverage is considerably lower. 25 In 1990, the baseline year of the Millennium Development Goals (MDGs), the global monetary poverty rate (using the international-dollar-per-day pover- ty line) was 36%. Under the MDGs, the world agreed on the target to halve this number by 2015. As it turned out, poverty actually fell to 10%, surpassing the original committment. It is worth noting that global monetary poverty was never higher than 50% in the period of systematic global measurement. In 1981, when it was first measured, the rate was 42%. Survey coverage was just 51% of the world population, and four regions (including Sub-Saharan Africa) had a survey coverage of less than 40%. For more details, please visit the Povcalnet website at http://iresearch.worldbank.org/PovcalNet/povDupli- cateWB.aspx. 26 If numeracy (rather than literacy) were used to calculate the learning poverty rate, the pattern might be expected to be different. Among adolescents, boys slightly outperform girls on PISA math scores, for example. Yet on the 4th-grade TIMSS math assessment, girls outperform boys in many countries, such as Saudi Arabia, Jordan, South Africa, and Indonesia, and in the median country there is no gender difference. 27 This analysis examines spells of improvement for each country within a given assessment, for the subsample of countries in which the same assessment was applied more than once between 2000 and 2018. 28 This message is consistent with the concerns about slow progress raised by UNESCO (2019), a report for the 2019 UN High-Level Political Forum on the SDGs. 29 See also Pritchett 2013 for an earlier discussion of this issue. 30 Pritchett 2013, World Bank 2018a. 31 An alternative assumption for this high scenario would be to set the rate of improvement for every country equal to the global 80th percentile. It turns out, however, that the regional 80th-percentile scenario yields faster global reductions in learning poverty. This is because the regional 80th percentile is higher in Sub-Saharan Africa than in other regions, and the regional scenario applies that more rapid rate to Africa’s large child population. 32 We are grateful to Simeon Djankov and Ambar Narayan for these points. Thanks to these properties, as well as the greater availability of data today, the Learning Poverty measure could serve the purpose once envisioned for a “Millennium Learning Goal” (Filmer, Hasan, and Prichett 2006). 33 Crouch and Rolleston 2017; Crouch and Gustafsson 2018. 34 Saavedra 2019a. 35 Guadalupe and others 2017. 36 Castles, Rastle, and Nation 2018. 37 Beteille and Evans 2019. 38 World Bank 2018b. 39 Beteille and Evans 2019. 40 Kotze, Fleisch, and Taylor 2019. 41 Piper 2018. 42 Banerjee and others 2017. 43 JPAL 2019 44 Banerjee and others 2017. 45 Kucirkova and others 2017. 46 Gove and Wetterberg 2011. 47 Anderson, Wilson, and Fielding 1988. Ending Learning Poverty: What will it take? | Endnotes 47 48 Gove and Cvelich 2011. 49 The Education Commission 2016. 50 Crabbe, Nyingi, and Abadzi 2014. 51 Piper and others 2018. 52 Piper and others 2018. 53 Fuje and Tandon 2016. 54 Piper 2010. 55 Friedlander 2013. Environmental factors associated with early reading achievement in the developing word: A cross-national study. The study found that up to 15% of the variance in student outcomes in the Philippines, Uganda, Mali, and Ethiopia were attributable to home environment factors. 56 Results for Development 2016. 57 Results for Development 2016. 58 For more information on the Book Chain, refer to the work of the Global Book Alliance (GBA). The GBA has created various analytical tools to help coun- tries strengthen their “book chains” to increase text access and use by young readers. 59 Determined through a Home Background Index, which uses indicators such as whether or not the child is of an ethnic minority, has a college-educated mother, a college-educated father, a computer at home, and/or Internet at home. 60 Rolleston, James, and Duc 2013. 61 World Bank 2018b. 62 Worldreader website. Accessed at: https://www.worldreader.org/learnings/. 63 Google and Bolo 2019; Koval-Saifi and Plass 2018. 64 USAID 2016. 65 Duc and Tam 2013. 66 Piper, Zuilkowski, and Ong’ele 2016. 67 Shin and others 2015; Taylor and von Fintel 2016. 68 Patrinos and Velez 2009. 69 TIMSS 2011. 70 USAID 2015. 71 Maurer 2010. 72 Macdonald and others 2017. 73 Results for Development 2016. 74 The technical terms for the relative complexity of a writing system is “orthographic depth.” Shallow orthographies are ones with simple letter-to-sound correspondences, often one-to-one. Deep orthographies by contrast have less direct relationships between letters and sounds. Pronunciation does not follow regular rules. Instead the same letter may be pronounced differently depending on the word itself, such as with “pint” and “mint” in English. In other cases, the exceptions group into regular rules, such as pronouncing “gh” with an “f” sound when it occurs at the end of a word in English. 75 Seymour, Aro, and Erskine 2003. 76 MoEYS 2018. 77 UNESCO 2018b. 78 Bulat and others 2017. 79 Shin and others 2015; Taylor and von Fintel 2016. 80 Xprize website, accessed at: https://www.xprize.org/prizes/global-learning; https://norad.no/en/front/thematic-areas/education/innovation/eduapp4syr- ia/positive-evaluation-findings-for-eduapp4syria/]. 81 Gove and Wetterberg 2011. 82 World Bank 2018e. 83 European Commission 2012. 84 We have used the Government Effectiveness indicator under the World Bank’s Worldwide Governance Indicators to discern, in a general manner, different institutional contexts for which potential constraints and suggested interventions may apply. The definition of Government Effectiveness relates to “cap- turing perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government’s commitment to such policies.” Accessed at https://info.worldbank.org/ governance/wgi/Home/Reports. 85 Piper and others 2018. 86 Beteille and Evans 2019. 87 Glewwe and others 2011. 88 Piper 2010. 89 World Bank 2018b. 90 This work is carried out with support from the Bill and Melinda Gates Foundation, the UK’s Department for International Development, and the govern- ment of Japan. 91 Only 40% of the children in Sub-Saharan Africa live in a country with a learning assessment with a strong design, implementation, and documentation. Endnotes | Ending Learning Poverty: What will it take? 48 PHOTO BY: © KHASAR SANDAG / WORLD BANK 49