Ending the Learning Crisis: A Plan, and Its Test in 2026

Introduction

Enrolment in primary and secondary school has grown by 327 million students since 2000, to 1.4 billion in 2024. Over the same period primary completion rose from 77 percent to 88 percent and upper-secondary completion from 37 percent to 61 percent, and Ethiopia, which enrolled 18 percent of primary-age children in 1974, enrolled 84 percent in 2024 (UNESCO).

The number of children and young people out of school has nonetheless risen for seven years in a row, and many of the children who do attend are not learning to read. Before the pandemic, 57 percent of ten-year-olds in low- and middle-income countries could not read and understand a simple text. In 2022 the World Bank, UNESCO, UNICEF and their partners estimated that school closures had pushed the figure to 70 percent, and they put the cost of that generation’s lost learning at $21 trillion in lifetime earnings (World Bank).

Education belongs among species-level problems because of the number of children involved and because schooling raises the incomes of the people who receive it. A review of more than 1,000 estimates from 139 countries by George Psacharopoulos and Harry Patrinos found that each additional year of schooling raises earnings by about 9 percent on average, and that the return for women is about 2 percentage points higher than for men (World Bank). Schools are also among the institutions most exposed to war and to cuts in foreign aid, which ties this plan to the series’ essays on war, starvation and poverty.

This essay sets out a plan to get every child into school and learning the basics, then tests the plan against the world of September 2026. Displacement and the rights of women and girls have their own essays in this series and appear here only where they bear directly on schooling.

Where things stand

The 2026 Global Education Monitoring (GEM) Report, published by UNESCO on 25 March 2026, counts 273 million children and young people out of school, about one in six of school age. The report names conflict and population growth, particularly in sub-Saharan Africa, as the main reasons that progress has slowed (UNESCO). Of the 273 million, 194 million are of secondary-school age. Roughly a third of young people worldwide do not complete secondary school, and on current trends the world will not reach 95 percent upper-secondary completion until 2105 (Human Rights Watch).

Measure	Figure	Source
Children and young people out of school	273 million (2026 GEM Report)	UNESCO
Of whom secondary-school age	194 million	Human Rights Watch, citing UNESCO
Ten-year-olds unable to read a simple text, low- and middle-income countries	70 percent (2022 simulation; 57 percent before the pandemic)	World Bank
Learning poverty, sub-Saharan Africa	89 percent (2022 simulation)	World Bank
Crisis-affected school-aged children	258 million in 87 countries	Education Cannot Wait
Of whom out of school	93 million	Education Cannot Wait
Additional teachers needed by 2030	44 million, 15 million of them in sub-Saharan Africa	UNESCO
Annual financing gap to reach national education targets by 2030	$97 billion	GEM Report

The learning poverty figures in the table are the least certain. The 70 percent estimate is a simulation of pandemic effects made in 2022, and the World Bank’s briefing on foundational learning, last updated in May 2025, still relies on it (World Bank).

Education Cannot Wait (ECW), the global fund for education in emergencies, estimated in June 2026 that crises disrupt the schooling of 258 million children, 21 million more than 18 months earlier; 74 million of the 93 million out of school live in the 20 most severe crises (Education Cannot Wait).

The GEM Report team now projects that aid to education will fall by 30 percent between its 2023 peak and 2027, faster than the 25 percent decline it forecast a year earlier. Aid to basic education, which covers pre-primary, primary and lower secondary schooling, fell by 15 percent between 2023 and 2024, twice the 8 percent fall in education aid overall (UNESCO GEM Report).

Part I: The plan

1. Measure learning in every country, every few years

Many low-income countries have no regular, comparable test of whether children in the early grades can read or do basic arithmetic, so their ministries cannot tell whether reforms are working. Under the plan, every ministry of education tests a representative sample of pupils in grades 2 to 4 at least every three years, reports the results against the UN’s minimum proficiency levels, and publishes them. The Coalition for Foundational Learning, whose members include the World Bank, UNICEF, UNESCO, the UK, USAID, the Gates Foundation and the Global Partnership for Education (GPE), would supply instruments and pay for countries that cannot; it has already committed to halving the share of children unable to read and understand a simple text by age ten (World Bank), a target it cannot track without these data.

Kenya’s Tusome programme and the reforms in the Brazilian state of Ceará, both described under element 2, relied on this kind of measurement. Tusome set national reading benchmarks in Kiswahili and English and communicated them down to each school (Journal of Educational Change), and Ceará paired literacy goals for every child with regular assessment used to guide teaching (GEEAP).

2. Teach the basics with structured pedagogy and targeted instruction

The Global Education Evidence Advisory Panel (GEEAP), co-hosted by the UK Foreign, Commonwealth and Development Office, UNICEF, USAID and the World Bank, published its “smart buys” report in May 2023. Drawing on over 550 evaluations, it named three “great buys”, approaches that are highly cost-effective and backed by strong evidence: giving families information on the benefits, costs and quality of education; supporting teachers with structured pedagogy; and grouping children for instruction by learning level instead of grade (GEEAP).

Structured pedagogy, in the panel’s definition, is a coordinated package of lesson plans, learning materials, skills-based teacher training and ongoing mentoring. It raised foundational literacy and numeracy at relatively low cost in Kenya, Liberia and South Africa and has been introduced nationwide in Kenya and at large scale in Uganda and several South African provinces (GEEAP). Tusome, the Kenyan national programme funded by USAID, scaled up earlier pilots through the government’s own system of county officers and classroom visits. Its external evaluation found effects of 0.6 to 1.0 standard deviations on English and Kiswahili outcomes. The same study found that officials used the classroom observation data less than they could have to target support where it was most needed (Journal of Educational Change).

Teaching at the Right Level, developed by the Indian NGO Pratham, groups children by what they can actually do for part of the school day. Six randomized evaluations across seven Indian states, by researchers including Abhijit Banerjee and Esther Duflo, found consistent gains. In Uttar Pradesh, short learning camps doubled the number of children who could read a paragraph or story, and the approach has since reached more than 60 million children in India and Africa (J-PAL).

Ceará shows what a whole system built around these ideas can achieve. The state has the fifth-lowest GDP per capita of Brazil’s 26 states, yet since 2005 it has recorded the largest increase in Brazil’s national education quality index in both primary and lower secondary school. Ten of its municipalities are in the national top 20, with Sobral in first place. Ceará tied part of the state’s transfers to municipalities to results and supported them toward universal literacy by the end of grade 2 (World Bank).

Under the plan, each government with high learning poverty adopts one national programme for grades 1 to 4 built on structured pedagogy and targeted instruction, taught in a language children understand, and runs it through the ministry’s own teachers, coaches and supervisors, as Kenya did with Tusome, instead of through a parallel donor project.

3. Get children into school and keep them there

Most out-of-school children are adolescents, and cost keeps many of them out. Primary schooling is free in law almost everywhere, but pre-primary and secondary schooling often are not, and households pay about a quarter of all education spending worldwide (UNESCO GEM Report). Governments under the plan abolish tuition and registration fees through lower secondary school. The GEEAP report adds cheap measures with good evidence behind them: reducing travel times to school and merit-based scholarships for disadvantaged children rank as “good buys”, and telling families about the returns to schooling ranks as a great buy (GEEAP).

School meals and cash transfers linked to attendance belong in the plan mainly on other grounds, since the same panel classes both as “effective but relatively expensive” ways to raise learning. School meals already operate in 84 percent of countries (UNESCO), and for many children they are the most reliable meal of the day, which links this element to the plan on starvation.

In 2025 countries set national benchmarks that would cut the number of out-of-school children by 165 million by 2030, to about 107 million (UNESCO). The plan has UNESCO publish each country’s progress against them every year.

4. Recruit, train and keep enough teachers

UNESCO estimates that the world needs 44 million additional primary and secondary teachers by 2030, 15 million of them in sub-Saharan Africa, at a cost of about $120 billion a year. The attrition rate among primary teachers almost doubled between 2015 and 2022, from 4.62 percent to 9.06 percent a year, and many leave within their first five years (UNESCO). The GEEAP report lists additional teachers to reduce class size, when nothing else changes, among its “bad buys” (GEEAP). The plan therefore ties recruitment to the structured materials and coaching in element 2, which help new teachers become effective sooner, and asks finance ministries in fragile states to pay teachers on time, since unpaid teachers leave.

5. Start before primary school

The GEEAP panel rates quality pre-primary education for children aged 3 to 5, and parent-directed early stimulation for children up to 36 months, as good buys (GEEAP). Only 60 percent of primary pupils worldwide attended at least a year of pre-primary education, and in sub-Saharan Africa fewer than one in four children of pre-primary age are enrolled (Human Rights Watch). The plan adds at least one free year of pre-primary education, attached to existing primary schools where possible to keep costs down.

6. Keep learning going through war and disaster

Children in crisis zones need schools protected from attack and from military use. The Safe Schools Declaration, a political commitment to that effect, had 123 endorsing states as of February 2026 (GCPEA). Education also needs funding from the first weeks of an emergency, through ECW and humanitarian appeals, for temporary learning spaces, catch-up classes and psychosocial support. Refugee and displaced children should be enrolled in host countries’ national schools, a policy the essay on displacement treats at length.

ECW’s record suggests the model works where it is funded. Between 2023 and 2025 its programmes reached 10.4 million children in 33 countries, 51 percent of them girls. Participation rose in 97 percent of multi-year programmes, and 96 percent reported gains in literacy or numeracy (Education Cannot Wait).

7. Pay for it, mostly from domestic budgets

UNESCO’s benchmarks ask governments to spend at least 4 to 6 percent of GDP, or 15 to 20 percent of public expenditure, on education (UNESCO). One country in three meets neither benchmark, and the gap between what countries spend and what their 2030 targets require is about $97 billion a year (UNESCO GEM Report). Domestic budgets will carry most of the cost. Aid matters most in the poorest and most fragile countries, where it pays for a large share of public education, and donors should concentrate it on basic education, on the programmes in element 2 and on crisis response. Debt relief for low-income countries belongs in the same discussion and is covered in the essay on poverty.

8. Use technology where evidence supports it

The GEEAP report lists investment in laptops, tablets and computers alone among its bad buys. It classes software that adapts to each child’s level, in schools that already have the hardware, as promising but with limited evidence on cost and on delivery at scale (GEEAP). The plan funds education technology only through randomized trials with independent evaluation before any national rollout, and treats devices as tools for the pedagogy in element 2.

Sequencing

In the first two years, the priorities are learning assessments in every country that lacks one, protection of existing crisis-education funding, and national foundational learning programmes in the countries with the highest learning poverty. Over two to five years, the aim is fee-free lower secondary schooling, a year of free pre-primary education, and teacher recruitment on the scale the 2030 benchmarks require. Universal completion of secondary school depends over the longer term on sustained domestic financing and on peace in the countries where most out-of-school children now live.

What makes it stick politically

Foundational learning programmes are more likely to survive changes of government when they belong to the ministry, when results are published for parents and legislators, and when teachers see the materials as help. Ceará has kept its reforms going for more than a decade, and its design gives municipal governments a financial stake in literacy results (World Bank). Fee abolition and school meals are popular with voters and tend to last once introduced. The risk is that enrolment rises faster than the supply of teachers and classrooms and quality falls, and regular publication of learning data gives reformers the evidence to resist that trade.

Where reasonable people disagree

Some economists argue that expanding access has done too little for learning and that scarce money should go almost entirely to quality in the early grades. Others point to the high returns to additional years of schooling, especially for women, and argue that secondary schooling cannot wait. Language of instruction is contested in multilingual countries, where teaching the early grades in children’s home languages means producing materials and training teachers in many languages at once. Low-cost private schools divide opinion between those who see them as a response to failing public systems and those who see them as a threat to universal free provision. There is also real uncertainty about artificial intelligence, whose first trial results are large but short-term and small in scale.

Part II: The plan in the world of September 2026

Measurement (element 1): no current count of learning

The 2026 GEM Report gives a clear count of exclusion, built from enrolment data and household surveys, but the count of children not learning is far weaker. The most widely cited figure, 70 percent learning poverty, remains a 2022 simulation, and the World Bank’s briefing still used it in 2025 (World Bank). The Coalition for Foundational Learning has a target of halving learning poverty but no current global measurement against which to judge progress, which makes element 1 more urgent than it was when the target was set.

Foundational learning (element 2): the evidence is shaping lending

Researchers at the Center for Global Development, led by Noam Angrist, published an analysis in March 2026 of 237 World Bank project appraisal documents and 72 UK business cases. Between 2015 and 2019, 12 percent of World Bank education projects mentioned structured pedagogy or targeted instruction. The share rose to 23 percent in 2020 to 2024 and to 68 percent in 2025, and both education projects the UK approved in 2025 mentioned both approaches (Center for Global Development). The authors caution that a mention does not guarantee serious implementation, but the shift in design is large, and it comes as the same agencies have less money to spend.

Access (element 3): some fees fall while funding for meals shrinks

Liberia announced on 5 August 2026 that it would abolish school registration fees from kindergarten through grade 6 from the start of the school year on 7 September. One-third of Liberian children have never attended school, and only 17 percent complete grade 9 (Human Rights Watch). The reform covers primary school, while the plan calls for the same step at lower secondary level, where most out-of-school children are.

Funding cuts are working against access. UNICEF warned in September 2025 that a projected $3.2 billion fall in education aid could push 6 million more children out of school by the end of 2026 and raise the global total from 272 million to 278 million. It expected donor funding for primary education to fall by about a third and support for school feeding to be cut by more than half, and it said 350,000 Rohingya refugee children risked losing access to basic education (UNICEF). The 2026 GEM count of 273 million relies on data that lag by a year or more and does not yet capture most of that effect.

Money (element 7): the heaviest blow to the plan

The GEM Report team calculates that total US aid fell by 57 percent in 2025, the largest reduction by any donor on record (the cuts are described in the essays on starvation and war). Germany has cut its development budget by a quarter since 2022, and the United Kingdom is reducing aid to 0.3 percent of national income by 2027. Some countries depend heavily on that money: aid paid for nearly 40 percent of public education spending in Rwanda over 2022 to 2024 and 31 percent in Sierra Leone (World Education Blog).

The pooled funds that finance the plan’s core are shrinking too. At an event co-hosted by Nigeria, Italy and the Global Partnership for Education (GPE) during the UN General Assembly on 23 September 2026, GPE announced $2.1 billion in confirmed and anticipated pledges from donor governments toward its $5 billion target for 2026 to 2030, plus $1.4 billion in co-financing from development finance institutions and other partners. Partner countries announced $471 billion in domestic education financing commitments, which GPE described as a 32 percent increase in spending by 2030 (TVC News). These totals come from press reporting of GPE’s announcement, because GPE’s own website could not be reached. The UK pledged £100 million, against £430 million in the previous cycle, a 77 percent cut (Save the Children UK). The Global Campaign for Education-US urged Washington on 16 September to make a strong pledge (GCE-US), and none of the published reporting on the event records a US pledge.

Humanitarian education funding has fallen for a second consecutive year and now covers less than a third of assessed needs. ECW is seeking $600 million for 2027 to 2030 to reach 10 million more children, at a replenishment conference hosted by Switzerland in Geneva on 5 November 2026 (Education Cannot Wait).

Conflict (element 6): attacks at record levels

The Global Coalition to Protect Education from Attack (GCPEA) published Education under Attack 2026 in June, reporting 8,566 attacks on education and cases of military use of schools in 2024 and 2025, more than 40 percent above the previous two years. The attacks harmed at least 10,600 students and education staff, and the highest numbers were recorded in Colombia, the Democratic Republic of the Congo, Ethiopia, Haiti, Palestine and Ukraine. Military use of schools rose to more than 1,900 cases from about 1,000, and GCPEA identified drone activity in about 300 attacks (Human Rights Watch). In a data update in September 2026, GCPEA raised the two-year total to 9,259, a 50 percent increase, after adding verified incidents for 16 countries. The revised count for Ukraine rose from 960 to 1,062 and for Haiti from 419 to 577 (GCPEA). The drone attacks fit the wider spread of drone warfare described in the essay on war.

Sudan. UNICEF reported on 4 May 2026 that at least 8 million children are out of school and that nearly half of all school buildings no longer function. A GPE-funded programme led by UNICEF and Save the Children aims to reopen about 850 schools in seven states and return more than 328,000 children to learning between 2026 and 2028 (UNICEF). The programme is element 6 in practice, but it reaches about 4 percent of the children out of school.

Gaza. Hundreds of thousands of children in Gaza returned to in-person classes on Saturday 19 September 2026 for the first time in nearly three years, in damaged buildings, makeshift schools and tents, Reuters reported. UNICEF says 98 percent of Gaza’s schools were damaged or destroyed. Local officials say about 660,000 students lost in-person schooling, at least 540,000 are expected to attend this year, and about 22,000 students and more than 770 education workers, 532 of them teachers, were killed (Reuters via Sowetan). On 24 September ECW committed $3 million to bring 10,000 to 13,000 children with conflict-related injuries and disabilities back to learning (Education Cannot Wait).

Iran and Lebanon. UNICEF reported that a missile strike on the Shajareh Tayyebeh School in Iran on the first day of the war killed 168 children, and that more than 760 schools in Iran had reportedly been destroyed or damaged by early April (UNICEF). In Lebanon, a ministry assessment supported by UNICEF counted 340 schools damaged by July, 17 of them destroyed, and UNICEF warned that at least 100,000 children could be left without a classroom when the 2026-27 school year begins (UNICEF Lebanon).

Afghanistan. Girls have been barred from secondary school since September 2021. UNICEF said in April 2026 that more than a million girls have been denied education as a result and that over two million will have been by 2030. The ban also shrinks the teaching force: the number of women teaching in basic education fell from nearly 73,000 in 2022 to about 66,000 in 2024, and UNICEF projects a loss of up to 20,000 women teachers by 2030 (UNICEF Afghanistan). The authorities also shut learning centres for girls above grade six during 2024 and 2025 (Human Rights Watch), and GCPEA’s September update raised Afghanistan’s count of attacks on education from 129 to 201 (GCPEA). The essay on women and girls covers the ban in detail.

Climate shocks close schools as well. On 15 September 2026 ECW committed $1 million for children in three districts of Nepal where floods damaged 21 schools and turned 8 more into shelters (Education Cannot Wait). The El Niño now under way, described in the essay on starvation, threatens harvests in southern Africa and with them the household incomes and school meals that keep many children in class.

Teachers and pre-primary (elements 4 and 5)

In 2026 the sharpest pressures on teachers and early education came from crises. Afghanistan is losing women teachers, Gaza reopened its schools after the deaths of 532 teachers, and ECW counts 70,000 of Gaza’s youngest children without access to preschool (Education Cannot Wait).

Technology (element 8): promising trials, thin evidence

The first rigorous trials of generative AI tutors in poor countries have produced results that justify further testing, though not yet scale-up. In a World Bank randomized trial in public secondary schools in Edo State, Nigeria, students used ChatGPT in supervised after-school sessions twice a week for six weeks, about 13 hours in total. They gained about 0.3 standard deviations overall and 0.24 in English, which the researchers equated to 1.5 to 2 years of typical learning and placed above 80 percent of rigorously evaluated education interventions. The authors said they could not isolate the chatbot’s effect from the prompts, curriculum alignment and teacher support around it, or say whether the gains last (World Bank).

In May 2026 Google reported a pre-registered, eight-week randomized trial in Sierra Leone, run with Fab AI and local teachers, covering 48 mathematics classrooms and nearly 1,800 students in grades 7 and 8. Students using Google’s Guided Learning tool improved by 0.26 standard deviations, and those who used it for 12 hours or more by 0.38 (Google). These figures come only from Google’s own blog post and technical report; the company makes the product tested, and the results had not appeared in a peer-reviewed journal.

Both trials used teacher supervision, curriculum-aligned content and a randomized design, as element 8 requires, but both were small and short and tested older pupils, while learning poverty is measured at age ten. AI tutoring for older pupils has earned larger and longer trials by independent researchers; for early-grade reading, structured pedagogy and targeted instruction have a decade of evidence behind them.

Law: a right to free pre-primary and secondary education

Formal UN talks on an optional protocol to the Convention on the Rights of the Child, which would recognize a right to free pre-primary and free public secondary education, began in 2025 (Human Rights Watch). The second formal session of the working group met in Geneva from 31 August to 2 September 2026, with 81 states attending, and Guatemala became the 62nd country to support the initiative publicly. The session did not agree a draft text: some states, including Chile and South Africa, backed moving to drafting, others, including Austria and Egypt, asked for more time, and governments agreed to hold informal consultations before the next formal meeting in October 2027 (Human Rights Watch; Human Rights Watch). A treaty would give families and courts a basis for challenging the fees that keep adolescents out of class, though it would not itself pay for schools.

Part III: Revised priorities for the next 6 to 12 months

Fund ECW’s replenishment in full at the Geneva conference on 5 November 2026, and protect humanitarian education funding for Sudan, Gaza, Lebanon, Afghanistan and the Rohingya response, where children are losing their schooling now.
Close the gap between GPE’s $2.1 billion in pledges and its $5 billion target, seek a pledge from the United States, which does not appear in published reporting of the September event, and direct new money to foundational learning programmes in the countries with the highest learning poverty.
Protect school meals from the aid cuts as both a nutrition and an attendance measure, with the El Niño response in southern Africa as the first test.
Fund learning assessments in every low-income country without one, so that a new global learning poverty estimate can replace the 2022 simulation.
Press the parties to the wars in Sudan, the Middle East and Ukraine to stop attacks on schools and military use of school buildings, and add endorsements to the Safe Schools Declaration.
Commission independent, larger and longer trials of AI tutoring in low-income countries before any national rollout, with results published whatever they show.
Use the consultations before the October 2027 session to settle the open questions on the optional protocol, so that drafting can begin then.

Conclusion

The plan tested in this essay depends on evidence about what works and on money to apply it. During 2026 the evidence grew stronger: structured pedagogy and targeted instruction have worked at national scale in Kenya and across a Brazilian state and now appear in most new World Bank education projects. The funding that would put them into classrooms is falling faster than forecast: aid to education is projected to fall 30 percent from its 2023 peak by 2027, and the UK has cut its GPE pledge by more than three-quarters.

Recorded attacks on education rose by half in 2024 and 2025, Sudan has at least 8 million children out of school, and Gaza’s children returned in September to tents and damaged buildings after nearly three years without classrooms. Where access returns, schooling can restart quickly, the same lesson the essay on starvation draws from the reversal of famine in Gaza.

The question for 2027 is whether governments and donors fund what the evidence supports before the children now out of school grow too old for catch-up to be realistic. Figures in this essay reflect reporting available in late September 2026, and several, particularly those on the GPE and ECW replenishments, will change within weeks.