The Education–Climate Vulnerability Trap

The first link is disruption. UNICEF counted more than 171 million students — about one in ten from pre-primary to upper secondary — whose schooling was disrupted by climate hazards in 2025: 109 million by storms, at least 70 million by floods and about 50 million by heatwaves, three-quarters of them in low- and lower-middle-income countries. Its previous count, with a different method, put the figure for 2024 at nearly 250 million. The World Bank counted 404 million students affected by climate-related school closures between January 2022 and June 2024, with schools in low-income countries losing an average of 18 days a year against 2.4 in richer countries.

The second link is learning on the days that remain. Park, Goodman, Hurwitz and Smith, following 10 million American students who retook the same exam, found that without air conditioning a 1 °F hotter school year reduced that year's learning by about one per cent, and that air conditioning largely offset the effect. Park, Behrer and Goodman extended the analysis to 58 countries and found that learning falls as hot school days increase, with the poorest students hit hardest. The World Bank estimates that students in Brazil's poorest municipalities could lose half a year of learning to heat alone.

The third and fourth links run back from learning to vulnerability. Across countries, adults' average years of schooling are strongly correlated with INFORM's measure of lack of coping capacity (r = −0.81). Within countries over time, an earlier report in this series found that each additional year of adult schooling has been associated with about a third fewer disaster deaths per thousand people affected. Lost learning today is therefore lost resilience tomorrow — which is why the trap compounds over generations.

Learner-heat-days multiply the number of children aged 5 to 19 in each country by the number of days a year on which the heat index reaches 35 °C where they live. Today the world's 1.94 billion school-age children experience about 44 billion such child-days a year. By 2040–2059, under the SSP2-4.5 climate and SSP2 population, the total reaches 91 billion; under the fragmented, high-emission SSP3-7.0 path with its larger population, 117 billion. Even with SSP1's smaller population, it would reach 72 billion. Days above 41 °C — the "danger" level — rise from 5 to 17 billion child-days.

The burden is shifting. South Asia bears 58% of learner-heat-days today and 39% by mid-century; sub-Saharan Africa's share rises from 7% to 24%, as its school-age population grows from 422 to 666 million and the number of hot days in West and Central Africa multiplies. Nigeria goes from 11 to 63 hot days a year, Bangladesh from 75 to 145, India from 58 to 102 and Pakistan from 84 to 116. Of today's school-age children, about 600 million live in countries averaging 30 or more hot days a year; on the mid-century climate, that would be about 1.1 billion.

Lower-middle-income countries carry most of the exposure: 79% of today's learner-heat-days and 69% by mid-century. Low-income countries carry only 4% today, because humid heat at this threshold is still uncommon across much of the Sahel's dry interior and the East African highlands, but their share almost triples, to 11%, and their schools are least able to respond. High-income countries account for 1–2%.

Learning-adjusted years of schooling, from the World Bank's Human Capital Index, combine how long children stay in school with how much they learn. Among countries at the same income, each additional ten days a year with a heat index of 35 °C or more is associated with about 0.17 fewer learning-adjusted years (t = −3.8), and about 0.11 fewer when each region is compared only with itself (t = −2.4). Harmonised test scores show a similar association with income held constant — about 4 points per 10 hot days — but it is not statistically distinguishable from zero within regions.

These cross-country patterns cannot prove that heat causes lower learning: hot countries differ in history, institutions and school systems in ways income does not capture. But they are consistent with the causal studies, which isolate heat by comparing the same students in hotter and cooler years. Those studies also explain why the effect is unequal. Air conditioning largely neutralised heat's effect on American students, and the effect was concentrated among students in schools without it — disproportionately poor and minority students.

The implication for the countries in this dashboard is sobering. Where most schools have no electricity, the remedy that protected American students is unavailable; where classrooms are crowded, poorly ventilated and roofed with sheet metal, indoor temperatures can exceed outdoor ones. The same heat that costs a well-equipped school little can cost an unequipped one much more.

School closures are the most visible way climate disrupts education. In 2025, storms disrupted schooling for 109 million students, floods for at least 70 million and heatwaves for about 50 million, across 106 countries and territories analysed by UNICEF. Closures are not evenly spread: low-income countries lose an average of 18 school days a year to climate-related closures, against 2.4 in high-income countries, according to the World Bank.

Some disruptions last far longer than the event. Pakistan's 2022 floods disrupted the education of an estimated 3.5 million children, and the World Bank estimated that a million of them might not return to school. Damaged buildings, schools used as shelters, displaced families and lost household income all extend closures, and each week out of school compounds into lost learning: the pandemic showed how quickly closures raise learning poverty, which rose from 57% to an estimated 70% of 10-year-olds in low- and middle-income countries.

Learning poverty is already highest where closures are longest. In sub-Saharan Africa an estimated 89% of 10-year-olds could not read a simple text in 2022; in low-income countries, 91%. Every climate-related closure there falls on children who are already far behind, and the evidence from the pandemic suggests that losses are not recovered without deliberate catch-up programmes.

Adapting schools to heat starts with electricity: fans, ventilation, cooling, lighting for shifted timetables and the devices for remote learning all need it. Weighted by the number of children, only 19% of primary schools in low-income countries have electricity, 40% basic drinking water and 19% basic handwashing facilities. In lower-middle-income countries the shares are 61%, 80% and 75%; in upper-middle-income countries 97%, 86% and 91%; and in high-income countries essentially all.

The gap is widest exactly where heat is rising fastest. In twelve countries projected to average 60 or more hot days a year by mid-century, fewer than half of primary schools have electricity — Chad (3%), Niger (7%), Mali (16%), Burkina Faso (21%), Togo (22%), Ghana (25%), Myanmar (27%), Benin (28%), the Gambia (34%), Bangladesh (43%), Mauritania and Senegal (44%) — and they are home to 121 million school-age children. Across all 30 countries with data where fewer than half of schools are electrified, 270 million children face a projected 17.5 billion learner-heat-days a year by mid-century.

Infrastructure is not the only capacity that matters. Teachers trained to adjust lessons and timetables, school leaders with authority to act on heat warnings, ministries with plans to keep learning going during closures, and families able to keep children in school after a disaster all count. But a school without power, water or shade has few options on a 40 °C day — and the cheapest fixes, solar power and passive cooling, are now within reach.

A country is placed in the trap when most of its 10-year-olds cannot read a simple text — learning poverty of 50% or more, or, where that is not measured, a harmonised test score below 380 — and it faces high climate hazard: an average of 60 or more days a year with a heat index of 35 °C by mid-century, or a high INFORM score (7 or more out of 10) for floods, cyclones or drought. Thirty-five countries meet both conditions. They hold 962 million school-age children, half the world's total: 426 million in South Asia, 242 million in sub-Saharan Africa, 142 million in the Middle East, North Africa, Afghanistan and Pakistan, 123 million in East Asia and the Pacific and 29 million in Latin America.

India, Pakistan, Nigeria, Indonesia, Bangladesh, the DR Congo, the Philippines and Egypt account for most of these children. Eight countries are in the trap because of heat alone, fifteen because of floods, cyclones or drought alone, and twelve because of both. On average fewer of their primary schools are electrified (63%, against 85% in other countries with data), their learning poverty is 65%, and their children will face 78 hot days a year by mid-century, against 38 today.

The trap's size in 2050 depends on the development path. Under SSP2 its school-age population rises to 1.05 billion; under SSP3, with higher fertility and slower progress, to 1.27 billion. The capacity side diverges too: adults in today's trap countries average 6.5 years of schooling, and by 2050 that reaches 10.6 years under SSP1, 9.6 under SSP2 and only 8.9 under SSP3 — a gap that will shape how well these societies cope with the climate of the second half of the century.

The trap layer marks countries in the trap in red, those with low learning but lower climate hazard in purple, and those with high hazard but lower learning poverty in amber. The heat layers show how the zone of frequent heat spreads from South Asia and the Gulf across West and Central Africa and South-East Asia.

The learning-poverty and school-electricity layers show the capacity side, concentrated in the Sahel and Central Africa. The growth layer shows where school-age populations will rise most by 2050 under SSP2 — overwhelmingly in sub-Saharan Africa — and where they will shrink.

Hover over a country for its card: school-age population now and in 2050, hot days now and by mid-century (shown as answer bubbles of ten days each), learning poverty and test scores, out-of-school rates, school electricity, water and handwashing, climate hazard and adults' schooling.

The first priority is schools that can function in heat. Passive measures — reflective and insulated roofs, cross-ventilation, shade trees and verandas, orientation away from the afternoon sun — lower classroom temperatures cheaply in new and existing buildings. Off-grid solar can power fans and lights in the four-fifths of low-income-country schools that have no electricity, and it also enables the digital tools that keep learning going during closures. Drinking water and handwashing protect children's health in heat and after floods. The World Bank estimates that a one-time investment of about $18.51 per child could mitigate much of the impact of climate shocks on schooling.

The second is protecting learning time. Heat-adjusted calendars and timetables — earlier starts, longer breaks, school holidays in the hottest months — avoid the worst hours. Early warning linked to education ministries lets schools prepare rather than close abruptly; building schools to withstand floods and storms and avoiding their use as long-term shelters shortens closures; and plans for remote and community-based learning keep children engaged when buildings are shut. After any closure, rapid assessment and targeted catch-up — teaching at the right level, structured pedagogy, tutoring — recover losses that otherwise persist.

The third is to recognise foundational learning as climate adaptation. The capacity to understand warnings, protect health, adopt new farming practices and move into less climate-exposed work rests on literacy, numeracy and schooling, and the evidence links each additional year of schooling to fewer disaster deaths. Keeping girls in school is especially powerful, because it also lowers fertility and child mortality. Climate finance that counts school resilience and foundational learning as adaptation — rather than leaving them to strained education budgets — would put money where the trap binds.

Coverage: 156 countries with at least 1 million people. Heat: days a year with daily maximum heat index ≥ 35 °C (and ≥ 41 °C), World Bank CCKP CMIP6 0.25° bias-corrected 26-model ensemble medians for 1995–2014 and 2040–2059 (SSP2-4.5 and SSP3-7.0), population-weighted using CCKP's SSP2 population grids for 2020 and 2050. School-age population: ages 5–19 from the Wittgenstein Centre Human Capital Data Explorer v3.3 (SSP1–3) for 2020 and 2050. Learner-heat-days = school-age population × hot days. Adults' mean years of schooling (age 25+): Wittgenstein Centre, by SSP.

Learning: learning poverty (World Bank, latest); harmonised test scores and learning-adjusted years of schooling (Human Capital Index, World Bank). School infrastructure: proportion of primary schools with electricity, basic drinking water and basic handwashing (UNESCO UIS via World Bank EdStats, latest to 2020). Hazard and coping: INFORM Risk 2026 river flood, coastal flood, tropical cyclone and drought scores (highest of the four) and lack of coping capacity. Partial associations: OLS with log GDP per head (PPP) and, where stated, World Bank region fixed effects; robust t-statistics. Trap: learning poverty ≥ 50% (or test score < 380 where learning poverty is missing) and either ≥ 60 hot days by 2040–2059 or INFORM flood/cyclone/drought ≥ 7.

Caveats: hot days are annual and not limited to the school year or to school hours; the heat index ignores shade and indoor conditions; infrastructure data are incomplete and dated; cross-country associations are not causal; the trap thresholds are judgements; UNICEF's disruption counts changed method between years. Related reports: Human Survivability Under the Global Polycrisis, heat and labour productivity and The Geography of Human Vulnerability Through 2100.