An estimated seven in ten 10-year-olds in low- and middle-income countries cannot read and understand a short, simple text, even though most of them are enrolled in school.
Written from the sources listed below, not independently peer-reviewed.
Officials in 35 low- and middle-income countries estimated that 47% of children in their country could read a simple text by the end of primary school. What was the actual average?
About 70%
10-year-olds in learning poverty in low- and middle-income countries
Simulated estimate for 2022, up from 57% in 2019, based on pre-pandemic assessments and the length of school closures. No newer global estimate has been published.
Data year 2022. Source: World Bank, UNESCO, UNICEF, FCDO, USAID and Bill & Melinda Gates Foundation, 2022
58%
Pupils worldwide reaching minimum reading proficiency at the end of primary school
Latest available global estimate, which refers to 2019; the other 42% fall short of the minimum level.
Data year 2019. Source: United Nations Statistics Division, 2026
680 million
Children whose learning levels are not counted
UNESCO Institute for Statistics estimate, reflecting countries that do not report learning outcomes to the SDG database.
Data year 2024. Source: UNESCO Institute for Statistics, World Education Blog, 2024
World Bank and UNESCO estimates of learning poverty. The 2015 and 2019 values are based on learning assessments covering most of the school-age population of these countries; the 2022 value is a simulation of the effect of pandemic school closures, made before new comparable test data were available.
When learning poverty was launched in 2019, the World Bank aimed to halve the 2015 rate of 53% to about 26% by 2030. Countries' own national targets, added together, would bring it down to 42% by 2030, still far from the SDG 4.1 aim that all children learn. Even that would require cutting the estimated 2022 level of about 70% nearly in half within eight years.
In low-income countries about nine in ten children were in learning poverty in 2019, and simulations suggest this did not improve by 2022.
Source: World Bank, UNESCO, UNICEF, FCDO, USAID and Bill & Melinda Gates Foundation, 2022Sub-Saharan Africa has the highest rate: an estimated 86% in 2019 and 89% in 2022. The largest pandemic increases were simulated for Latin America and the Caribbean (52% to 79%) and South Asia (60% to 78%), where schools stayed closed longest.
Source: World Bank, UNESCO, UNICEF, FCDO, USAID and Bill & Melinda Gates Foundation, 2022Early grade reading assessments of more than 500,000 pupils in 48 countries found that over 90% could not name letters or their sounds, or read simple words, at the expected level after three years of school.
Source: World Bank press release on a Global Education Evidence Advisory Panel report, 2025In reading, boys are usually the ones left behind: only about a third of countries with data reach gender parity at the end of primary school, and in nearly all of the others boys score lower than girls.
Source: United Nations Statistics Division, 2026Each country is shaded by its latest available value, so you can see where the gap is widest. Hover or tap a country, search for one, or open the full table.
Loading country values…
Learning poverty: the share of children at the end of primary school age who are out of school or read below the minimum proficiency level. Latest year available per country since 2011, so years differ (most values are from 2019 or 2021, before or during pandemic disruption). Countries use different national, regional and international assessments, and many low-income countries have no recent data, so blank countries are not problem-free.
Source: World Bank and UNESCO Institute for Statistics, via World Bank Open Data. Latest year available per country; years differ.
Values load from World Bank Open Data. Simplified Natural Earth borders; they imply no position on disputed territories.
Instead of one “percent solved” score, DO GOOD rates three separate dimensions. A problem can have mature solutions that still reach too few people. These ratings are an editorial reading of the sources on this page, not a measured index.
Has the situation improved in the real world?
LowLearning poverty in low- and middle-income countries edged up from 53% in 2015 to 57% in 2019, before the pandemic, and was estimated at about 70% in 2022. In low-income countries about nine in ten children are affected.
How mature and validated are the available solutions?
HighStructured pedagogy and teaching at the right level are rated among the most cost-effective education investments, and a national evaluation in Zambia shows that teaching basics by level works when run by government teachers. Results still depend on how well programmes are implemented.
How widely are effective solutions actually used?
LowProven approaches reach only a minority of children: early grade assessments in 48 countries found that more than 90% of pupils could not read simple words at the expected level after three years of school.
The most recent global estimate is a simulation, the latest comparable global reading figure refers to 2019, and UNESCO estimates that the learning levels of 680 million children are not counted.
Learning poverty was already creeping up before 2020 and is estimated to have risen sharply with school closures; no global data yet show a recovery to pre-pandemic levels.
Learning to read with understanding by about age 10 is the foundation for almost everything a child learns later in school. The World Bank and UNESCO call the failure to reach this milestone learning poverty. Before the COVID-19 pandemic, 57% of 10-year-olds in low- and middle-income countries were already in learning poverty, and simulations suggest the share rose to about 70% in 2022 after long school closures. Many of these children are in school: the problem is not only that some are absent but that the years others spend in class do not teach them the basics. This page focuses on what children learn; whether they get into school at all is covered on the education access page, and the supply of qualified teachers on the teacher shortage page.
The burden falls hardest on the poorest countries. About nine in ten children in low-income countries were in learning poverty in 2019, and in sub-Saharan Africa the estimated share was close to nine in ten in 2022. Children who do not learn to read early tend to fall further behind each year, repeat grades and leave school early, because later lessons assume they can read. The World Bank estimates that the learning lost during the pandemic alone could cost today's students US$21 trillion in lifetime earnings. Because many countries rarely or never test early reading, the problem often stays hidden from the officials who could act on it.
SDG target 4.1 asks that all girls and boys complete primary and secondary school with relevant and effective learning outcomes. Learning poverty, developed by the World Bank and the UNESCO Institute for Statistics, condenses this into one indicator for the end of primary school:
Learning deprivation
The share of pupils at the end of primary school who read below the minimum proficiency level set for SDG indicator 4.1.1, as measured by national, regional or international assessments.
Schooling deprivation
The share of children of primary school age who are not in school. All of them are assumed not to read at the minimum level.
Combining the two
Learning poverty equals the out-of-school share plus the share of enrolled children below minimum proficiency, weighted by the share in school. The result approximates the share of 10-year-olds who cannot read and understand a short, age-appropriate text.
Related SDG measures
SDG indicator 4.1.1 tracks the share of pupils reaching minimum proficiency in reading and mathematics in grades 2 or 3, at the end of primary and at the end of lower secondary school. Data for the early grades are the scarcest.
What works
Teachers receive a coherent package of daily lesson plans, matching reading and maths materials for pupils, and regular coaching. An evidence panel convened by the World Bank, the UK government and UNICEF rates it as a top cost-effective way to raise foundational learning in low- and middle-income countries.
Limits
It works only if materials arrive on time, lessons are taught in a language children understand, and coaching continues after donor projects end.
What works
Children are assessed with a simple test and grouped by what they can do rather than by grade for part of the school day. In Zambia the Catch Up programme reaches more than one million pupils a year in over 5,000 schools; a randomised evaluation in more than 1,100 schools found gains in basic literacy and maths and better results on national exams at the end of primary school.
Limits
Short-run gains were limited to the skills taught, and adding teacher training to the Zambian programme doubled the cost per pupil without extra learning gains. It needs protected time in the timetable and backing from education authorities.
What works
A 2025 review of about 120 studies covering more than 170 languages finds that children learn to read best when teachers teach core skills step by step, including systematic phonics (the link between letters and sounds), and give pupils enough time to practise reading and access to books.
Limits
Teachers need structured guidance, easy-to-use materials and ongoing professional development to teach this way, which takes sustained funding. Decoding skills are a starting point; understanding what is read needs continued attention in later grades.
What works
The 2025 reading review recommends choosing languages of instruction that suit the children in the classroom and giving them explicit support to learn to read in each language they are taught in.
Limits
In countries with many home languages, producing books, training teachers and planning the transition to a national or international language takes time, money and political agreement.
What works
Assessing learning regularly is one of the five elements of the World Bank's framework for learning recovery, because it shows governments and teachers how many children can actually read. In one survey, officials in 35 low- and middle-income countries estimated that 47% of children could read, while the actual average was 25%.
Limits
Measurement only helps if results reach teachers and lead to changes in teaching. Large-scale assessments cost money, and comparisons between national and international tests are not always straightforward.
Promising, but not proven at scale. The stage shows how far each approach has come; it is not a recommendation.
Software that adjusts exercises to each child's level has shown large effects in some rigorous studies. The evidence panel classes it as promising but notes that evidence on cost-effectiveness and on use at scale is lacking, and that it only makes sense where devices already exist in schools.
Tutoring and learning messages delivered to families by phone have been tested in several countries. The evidence panel rates mobile phones for learning as promising, with limited evidence so far.
Results from Zambia suggest that a simple focus on basic skills, without costly extra training, can deliver most of the gains. Simpler designs are cheaper to run nationally, but which elements can safely be dropped in other countries is not yet known.
Researchers in Zambia analysed 4.4 million national exam scores over 12 years to measure long-term effects, showing that short tests of targeted skills can understate or overstate a programme's value. The approach depends on reliable administrative data, which many countries lack.
Knowledge gap
UNESCO estimates that the learning levels of 680 million children are not counted. Since 2015, no reading results at the end of primary or lower secondary school have been reported for 93% of children in Central and Southern Asia and 62% in sub-Saharan Africa and Eastern and South-Eastern Asia.
Coordination gap
In a survey of policymakers in 35 low- and middle-income countries, 81% agreed there was a learning crisis, but they greatly overestimated how many children could read in their own country, which weakens the pressure to act.
Implementation gap
The 2025 reading review names the failure to use evidence-based reading instruction as a main cause of the crisis. Teacher training and textbooks often do not teach reading explicitly and step by step.
Implementation gap
Extra textbooks, classrooms, smaller classes or grants alone, without changes in teaching, have repeatedly been shown not to be cost-effective for improving learning, because they leave the way children are taught unchanged.
Technology gap
Buying laptops, tablets or computers on their own is classed as a poor investment by the evidence panel, while software that adapts to a child's level is still unproven at scale and only an option where schools already have devices.
Coordination gap
When lessons follow the grade curriculum regardless of what children have learned, pupils who miss the basics can fall further behind each year. Changing curricula, exams and timetables to make room for catch-up requires agreement across many parts of an education system.
Unresolved questions, not vetted breakthroughs. Answering any of them would change what is possible.
The 2022 estimate of about 70% is a simulation. New comparable assessments are needed to show how far children have actually fallen behind or caught up, and in which countries.
Which elements of structured pedagogy and teaching at the right level are essential when governments run them for every school, and which can be dropped to lower cost without losing results?
How should countries with many home languages sequence reading instruction and the move to a national or international language, and how can good reading materials be produced cheaply in each language?
How can countries assess early reading regularly and cheaply while keeping results comparable across countries and over time, especially in the early grades where data are scarcest?
When donating to education, choose organisations that measure whether children can read and do basic maths and publish those results, ideally working with public schools.
Reading aloud with children and giving them access to books in a language they understand helps build early reading skills. Volunteer reading or tutoring programmes run by schools or libraries welcome help.
Ask your elected representatives whether education aid from your country goes to approaches with strong evidence, such as structured teaching and regular learning assessment.
Ideas visitors sent in for this challenge, read by the DO GOOD team and pinned with permission. The colour shows the kind of contribution.
Loading the board…
What happens next
The DO GOOD team reads every idea and sorts it by colour: solution, research question, existing project, evidence or local action.
If you agree to publication, it can be pinned to this board. Contact details are never shown.
Pinned ideas are not reviewed by scientists and are not endorsed by DO GOOD.
This page is a research overview. DO GOOD does not run a dedicated programme on this challenge, and donations are not earmarked for it. You can join our volunteer work and outreach.
World Bank, UNESCO, UNICEF, FCDO, USAID and Bill & Melinda Gates Foundation, published 2022
Most recent global learning poverty estimate. Source of the 53% (2015), 57% (2019) and simulated 70% (2022) rates for low- and middle-income countries, regional values, the 9 in 10 in low-income countries, the 42% implied by national targets, the US$21 trillion earnings loss, the survey of officials in 35 countries, and the indicator definition.
United Nations Statistics Division (data from the UNESCO Institute for Statistics), published 2026
Official SDG progress report. Source of the latest global estimate of 58% of pupils reaching minimum reading proficiency at the end of primary school (2019), gender parity in reading, and gaps in early-grade learning data.
World Bank press release on a Global Education Evidence Advisory Panel report (FCDO, UNICEF, World Bank), 30 October 2025, published 2025
Review of about 120 studies in more than 170 languages on how to teach reading. Source of the early grade reading results for more than 500,000 pupils in 48 countries and the recommendations on explicit instruction, phonics, language of instruction, practice time and books.
Global Education Evidence Advisory Panel (World Bank, FCDO, UNICEF Innocenti, USAID), published 2023
Expert review of more than 550 evaluations ranking education investments. Source of structured pedagogy and teaching by learning level as Great Buys, adaptive software and mobile phones as promising, and hardware or inputs alone as Bad Buys.
J-PAL (Abdul Latif Jameel Poverty Action Lab), 18 March 2026, published 2026
First randomised evaluation of a national scale-up of teaching at the right level (Catch Up, Zambia): more than 1,100 schools, over 5,000 schools and one million pupils a year reached, long-term effects from 4.4 million exam scores, and costs of added teacher training.
UNESCO Institute for Statistics, World Education Blog (GEM Report), 7 February 2024, published 2024
Source of the estimate that the learning levels of 680 million children are not counted, and of regional gaps in reported reading assessments since 2015.
DO GOOD International (2026). Ending Learning Poverty. Global Challenges. Updated October 15, 2026. https://www.dogood.world/challenges/learning-poverty
For figures, please cite the original sources listed above.