World Development Report 2018: Learning to Realize Education’s Promise

The global schooling expansion hides another statistic: for millions, schooling is not producing enough learning. Learning outcomes in basic education are so low, in so many contexts, that the developing world is facing a learning crisis. In many low-income countries, learning levels are low in an absolute sense, while in many middle-income countries average learning levels remain far behind those in high-income countries. The learning crisis disproportionately affects children from poor households: they are far more likely to leave school without acquiring basic skills like literacy or numeracy. Ultimately, the learning crisis translates into severe shortcomings in the skills of the workforce.

For too many, learning isn’t happening

In South Africa, 27 percent of 12-year-olds were enrolled in grade 6 but were functionally illiterate; in Zambia that share was 44 percent (Hungi and others 2010).

Globally, 125 million children are not acquiring functional literacy or numeracy, even after spending at least four years in school. In Malawi and Zambia in 2012, more than 89 percent of students could not read a single word by the end of grade 2. In Guyana in 2008 (when the most recent data were collected), that share was 29 percent at the start of grade 3. A similar picture emerges for numeracy. In rural India in 2016, less than 28 percent of students in grade 3 could master double-digit subtraction. Emerging data on student achievement show that, for millions, schooling is producing little learning in crucial early grades.

Millions complete primary education without acquiring the basic competencies needed for further learning. According to a 2014 regional assessment, among grade 6 students in West and Central Africa, nearly 58 percent are not sufficiently competent in reading or mathematics to continue schooling (figure 3.1). Similarly, the most recently available regional assessment of grade 6 students in southern and East Africa (from 2007) shows that 37 percent are not competent in reading, and more than 60 percent are not competent in mathematics (figure 3.2). This lack of basic competency is systematically lower for students from poorer families. In Honduras, half of the grade 6 students from the poorest quintile scored at the lowest reading competency level in a 2013 regional Latin American assessment; only 7 percent of those from the richest quintile did (figure 3.3).

Low-performing education systems are failing to meet their own curriculum standards (box 3.1).

While test scores on international assessments may

These low learning levels are not an inevitable be worryingly low, similar patterns emerge when by-product of rapidly expanding education. Starting assessing students against national standards. In in the 1950s, the Republic of Korea focused on ensururban Pakistan in 2015, only three-fifths of grade 3 ing quality primary education for the vast majority of students could correctly perform a subtraction like its population before shifting to a similar emphasis

54 – 25; in rural areas only two-fifths could. Across 51 on secondary and ultimately higher education—with countries, only about half of women who completed excellent learning results. The success of this strategy grade 6 (but no higher) could read a single sentence.

shows it is possible to ensure quality education even

It is hard to imagine that these women are reaping while rapidly expanding schooling. The key ingredithe full potential economic or social returns from ent is a persistent emphasis on the needs of the poor their years of schooling.

and disadvantaged. The more recent experience of Vietnam reinforces that lesson; it, too, has maintained education quality during rapid expansion by ensuring that disadvantaged students receive relatively equitable access to quality schooling.

Even in middle-income countries, millions of students are lagging behind. In Brazil, internationally comparable assessments reveal that more than three-quarters of youth are reaching the age of 15 without being able to perform at the lowest level of competence on the Programme for International Student Assessment (PISA) tests. Similarly, a third of students in Paraguay have only a basic grasp of reading skills (“reading for meaning”) by grade 6. These students are therefore ill-equipped to participate in their economy and society. These numbers also show interesting gender-based differences (box 3.2).

However, some countries are doing better. Albania, Peru, and Portugal have made impressive progress in improving average student achievement relative to countries with similar incomes. Latvia outperforms several countries in eastern Europe; Vietnam is a positive outlier in Southeast Asia. Although it is not always possible to clearly isolate the factors responsible for systemwide improvements in student learning, a policy focus on education quality appears to be important. For example, a major component of Vietnam’s strong performance has been a convergence in school quality within the country. The share of schools that meet the national standards of quality has steadily increased over the last 25 years.

Low student achievement in some middle-income countries relative to their economic competitors signals a failure to live up to their own expectations. According to the leading international assessments of literacy and numeracy, the average student in low-income countries performs worse than 95 percent of the students in Organisation for Economic Co-operation and Development (OECD) countries—meaning that student would be singled out for remedial attention in a class in a wealthier country. In Colombia, Indonesia, and Peru, student performance at the 75th percentile on the PISA math test is barely above that at the 25th percentile of the OECD average. In Algeria, the Dominican Republic, Kosovo, and Tunisia, it is below the 25th percentile of the OECD average (figure 3.4). The disparity between the average PISA score for Latin American countries and OECD countries is equivalent to over two full years of math education. Based on its rate of progress in average PISA scores from 2003 to 2015, it would take Tunisia over 180 years to reach the OECD average in math. This slow rate of improvement is especially problematic for middle-income countries trying to position themselves as important players in the global economic landscape.

The mapping between schooling and workforce skills varies dramatically across countries. For example, the working-age population in Colombia reaches basic literacy proficiency by the lower secondary percent of Ghana’s working-age population and over 60 level, whereas the population of Bolivia needs six percent of Kenya’s have just level 1 literacy or below— more years to attain even close to the same profithat is, their literacy proficiency is limited to underciency. Similarly, among 18-to 37-year-olds in Nigeria, standing basic texts, but they are not able to integrate, only 19 percent of primary completers can read; in evaluate, or interpret information from a variety of text

Tanzania, 80 percent can.

materials (figure 3.5). This contrasts with the average

In some countries, large proportions of “educated”

for high-income countries, where only 15 percent of the working adults are effectively low-skilled. Nearly 80 working-age population is at level 1 or below. Individuals underpin the development of higher-order cognitive, technical, and specialized skills. This skills deficit limits opportunities for further education or training because the capacity to make up for lost skills shrinks over time: second-chance adult education programs have limited success, and on-the-job training usually favors workers with more education and skills. The consequences are dead-end jobs with relatively flat lifetime income growth for students leaving school with poor foundational skills, a situation that will only get worse as technology affects the demand for skills (see spotlight 5).

Estimates based on 41 countries where skill measures are available suggest that, globally, more than 2.1 billion of 4.6 billion working-age adults (ages 15–64) lack crucial foundational skills. Among younger adults (ages 15–24), the number is 418 million. While these skills gaps exist in all countries, their magnitude is greater in developing countries (figure 3.6), with an estimated 92 million 15-to 24-year-olds affected in East Asia and Pacific, 120 million in South Asia, and 47 million in Latin America and the Caribbean. The implications, already profound, will be felt more acutely as jobs continue to shift from physical to more cognitive or socioemotional tasks. Progress in with low literacy proficiency are poorly prepared for the labor market, further education, and on-the-job training. In rapidly modernizing labor markets, most high-quality jobs—and even job training—require reading competency beyond minimum proficiency.

Low skills continue to undermine career opportunities—and earnings—long after students leave school. Gaps in foundational skills affect not only the starting points of new workers entering the labor market but also their growth trajectories. Good foundational skills are essential for further skills accumulation. Worldwide, many students leave school without mastering the key cognitive skills that meeting global development goals will be limited as evidence suggests not. In Pakistan, rich-poor learning long as the dimensions of this problem, its origins, gaps are smaller than learning gaps between children and its implications remain unrecognized.

from good and bad schools. In tests of English language, the difference in learning between a high-performing and a low-performing public school is 24

Poor children learn the least, which hurts them the most

times the difference between children from poor and nonpoor backgrounds, after controlling for observed child-level differences. Analysis of the 2009 PISA

Learning deficits are largest for poor people. In nearly found that “the best performing school systems [in all countries, students’ family backgrounds—includCanada; Finland; Hong Kong SAR, China; Japan; the ing parental education, socioeconomic status, and

Republic of Korea; and Shanghai, China] manage to conditions at home (such as access to books)—remain provide high-quality education to all students” rather the largest predictors of learning outcomes (figure than only to students from privileged groups.

3.7). In France, the difference in science performance on the 2015 PISA between the richest and the poorest

What is causing the learning crisis?

students was 115 points. In Hungary, this difference was 202 points. A 100-point difference in PISA scores is roughly equivalent to three years of schooling.

The learning gap between rich and poor students

A simple framework can be used to organize the proxgrows as students move to higher grades. In South imate (or immediate) determinants of the learning

Africa, children in grade 3 from the poorest households crisis. Proximate determinants are those most are three years’ worth of learning behind children from directly linked to learning outcomes and are themthe richest households. This gap grows to four years’

selves the result of deeper determinants. The frameworth of learning by grade 9. In Andhra Pradesh, work identifies four proximate determinants: learner

India, testing the same set of students each year reveals preparation, teacher skills and motivation, the that this gap increases every year after grade 2.

availability of relevant inputs, and the school manAre learning gaps between rich and poor students agement and governance that bring these together simply a matter of household characteristics? Recent (figure 3.8). This approach provides a simple tool for systematically integrating a wide range of actors and factors in determining the learning process.

Children from disadvantaged backgrounds tend to exhibit learning deficits years before they start school; these deficits leave them ill-prepared for the demands of formal education. Acquiring foundational skills in early childhood is essential for learning, and robust early childhood development can launch children on higher learning trajectories (see spotlight 1). But for children from disadvantaged backgrounds, adversities begin to accumulate before they are born. Chronic malnutrition, illness, the cumulative effects of material deprivation, low parental support, and the unpredictable, chaotic, or violent environments that can be associated with poverty all undermine early childhood development learning (see spotlight 2).

Steep socioeconomic gradients in cognitive, linguistic, and early literacy development help determine school outcomes. Language and cognitive gaps are evident before a child’s first birthday. In a wide range of countries, from the Democratic Republic of Congo to the United States, children from poor households lag behind their more affluent peers by age 3, with gaps widening as children age (figure 3.9). Gaps in early language and cognitive abilities are very alarming because they are important predictors of performance throughout school and into early adulthood.

Poor foundations are evident in other crucial determinants of school performance such as socioemotional and executive functions. Socioemotional skills include teamwork, motivation, and confidence, while executive functions (which rely on both socioemotional and cognitive skills) include planning, organizing, implementing, and multitasking, among others. The evidence on these developmental dimensions is more limited because of measurement difficulties. Still, gaps in working memory and sustained attention (executive functions) for poorer children are evident starting at 6 months of age and through the preschool years. Even in Madagascar, a very poor country, wealth gradients are apparent after accounting for maternal education and household inputs—gaps that widen with age. There are wealth gradients of socioemotional development as well. One in every three children between the ages of 3 and 4 in a range of countries fails to meet basic milestones in socioemotional development, such as the ability to control aggressive behaviors, avoid distractions, and get along with peers.

Because learning is cumulative and skills beget skills (see spotlight 1), the cognitive and socioemotional developmental gaps that emerge at young ages worsen over time. So do learning gaps: poor developmental foundations and lower preschool skills mean disadvantaged children arrive at school late and unprepared to benefit fully from learning opportunities. As these children get older, it becomes harder and harder for them to break out of lower learning trajectories.

Teachers are the most important determinant of student learning. Estimates suggest that in the United States, students with great teachers advance 1.5 grade levels or more over a single school year, whereas those with a poor teacher advance just 0.5 grade levels. Across kindergartens in Ecuador, differences in learning outcomes for language, math, and executive function are strongly associated with differences in teacher behaviors and practices. No other school-level factor has an impact nearly this large on student achievement.

But high-quality teachers are in short supply in low-income countries. Less than 25 percent of Sub-Saharan Africans currently complete secondary education. Thus there are simply not enough qualified candidates to meet the growing demand for teachers. In some countries, teacher training colleges have had to lower their entry requirements to ensure an adequate supply of teachers. These colleges also push trainees through the programs in two years or less to try to fill the need for primary school teachers. In Latin America, there is evidence that candidates entering the teaching profession are academically weaker than the pool of higher education students. Fifteen-year-olds who identified themselves as interested in a teaching career had much lower PISA math scores than students interested in engineering in every country in the region, and they scored below the national average in nearly all countries. Teachers may also not always have the necessary pedagogical skills; classroom observations in six countries in Sub-Saharan Africa found that few public primary school teachers are able to assess children’s abilities and evaluate students’ progress, and few engage in the practices typically associated with good teaching.

As a result, teachers often do not have sufficient mastery of concepts they are expected to teach. In several Sub-Saharan countries, the average teacher does not perform much better on reading tests than the highest-performing grade 6 students. Across six countries in the region, 40 percent of primary school teachers are not as knowledgeable as their students should be (table 3.1). In Bihar, India, only 10.5 percent of tested public school teachers are able to solve a

This problem is particularly concerning because three-digit by one-digit division problem and show the bulk of national education budgets goes to teacher the steps correctly.

salaries. In Latin America and the Caribbean, teacher

Many developing countries suffer significant salaries absorb nearly 4 percent of the regional losses of instructional time (figure 3.10). Unangross domestic product (GDP). Staffcompensation nounced visits to primary schools in six countries accounts for 80 percent of public spending on edufound that in public schools, on average, about one cation in some countries (figure 3.11). If one in five teacher in five was absent on a typical school day.

government primary school teachers is absent from

Even when teachers are present in school, they may school, developing countries are wasting considernot be teaching. In seven Sub-Saharan countries, able resources.

students receive only about two and a half hours of

According to recent data on 1,300 villages in teaching a day—less than half the scheduled time.

India, nearly 24 percent of teachers were absent

Teacher absenteeism and low time on task when in during unannounced visits, at an associated fiscal class—combined with other factors such as informal cost of US$1.5 billion a year. Reducing absenteeism school closures or student absenteeism—mean that in these schools would be over 10 times more cost-only about one-third of the total instructional time effective at increasing student-teacher contact time is used in Ethiopia, Ghana, and Guatemala. Even than hiring additional teachers.

in middle-income countries in Latin America, about 20 percent of potential instructional time is lost—the equivalent of one less day of instruction a week.

The effective management of schools relies on capacThere are many reasons for this loss of instructional ity and autonomy for decision making at the school time, including poor training and other demands on level, which are often lacking. Higher management teachers, and some teachers may perceive it as justiquality and school leadership are associated with fied (box 3.3). But whatever the cause, lost teaching better education outcomes. Yet in many developing time reduces student learning.

countries effective school management is missing (figure 3.12). Moreover, lack of autonomy prevents head teachers or school management committees from improving service delivery. Even when the requisite autonomy exists, it may not be enough. Schools may choose not to exercise the provided authority or may lack the will and capacity to do so. For example, a survey in Uganda found that only 57 percent of school management committee members reported having read their committee’s handbook. In Uttar Pradesh, India, a quarter of village education committee members surveyed did not even know they were members.

In many developing countries, the expansion of inputs has not kept pace with the explosion in enrollments. Governments have built classrooms and recruited teachers at unprecedented levels. But these efforts may not have kept up with rising enrollments, leading to a decline in per capita input availability. In Malawi between 2008 and 2015, as the gross enrollment rate in primary schools increased from 131 to 146 percent, the average number of students per class increased from 85 to 126. Uganda introduced universal primary education in 1997. The 68 percent increase in primary school enrollment that followed increased the student-teacher ratio from 38:1 in 1996 to 80:1 in 1997 and the student-to-classroom ratio from 68:1 in 1996 to 105:1 in 1997.

The learning crisis is real, but too often education systems operate as if it is not. Many policy makers do not realize how low learning levels are. Others do not acknowledge them or simply equate low learning with low resources. Still, there are reasons for optimism. First, learning is increasingly in the spotlight. Second, learning metrics are generating irrefutable evidence of the learning crisis, thereby creating pressure for action (see chapter 4). Third, promising new insights on how to tackle the crisis are becoming available (see parts III and IV of this Report).