Human Exposure Across the Polycrisis Century

In the recent climate, 3.4 billion people, 45% of humanity, live with at least one of three hazards: frequent humid heat (1.9 billion), extreme rain (1.5 billion) or long dry spells (1.0 billion); 1.0 billion live with two or more. All four pathways raise these numbers by the 2030s. By the 2090s, the total reaches 4.3 billion under SSP1-2.6 (58% of a smaller population), 6.2 billion under SSP2-4.5 (67%), 5.8 billion under SSP5-8.5 (74%) and 9.3 billion under SSP3-7.0 (76% of 12.2 billion).

Heat drives the growth. People living with frequent humid heat rise from 1.9 billion to 5.1 billion under SSP2-4.5 and 8.0 billion under SSP3-7.0, and those living with dangerous heat, a heat index above 41 °C on a month's worth of days a year, from 0.11 billion to 2.2 billion and 5.0 billion. Extreme rain and dry spells grow more slowly, by 55% and 73% under SSP2-4.5. Two or more hazards together rise from 1.0 billion to 2.8 billion under SSP2-4.5 and 4.7 billion under SSP3-7.0.

SSP1-2.6 is the only pathway in which exposure peaks: at 4.9 billion around 2050, falling to 4.3 billion by the 2090s as warming stabilises and population declines. SSP5-8.5 has the most heat per person but a smaller population than SSP3-7.0, so its totals sit between SSP2 and SSP3. The difference between the low and high pathways, 5 billion people by the 2090s, is the scale of what mitigation and demography decide.

Today, South Asia holds the largest exposed population, 1.4 billion people, followed by East Asia (0.9 billion), the Middle East and North Africa (0.55 billion) and sub-Saharan Africa (0.37 billion). By the 2090s under SSP2-4.5, South Asia's rises to 1.9 billion, but sub-Saharan Africa's quadruples to 1.6 billion, and its share of the world's exposed people goes from 11% to 25%. Under SSP3-7.0 it reaches 2.4 billion. East Asia (1.4 billion), the Middle East (0.9 billion) and Latin America (0.3 billion) also rise.

Two forces push exposure south. Sub-Saharan Africa's population grows fastest, and warming spreads frequent humid heat across West and Central Africa, which today have little of it: Nigeria's exposed share rises from 37% to 95%, almost all of it from heat. New exposure also appears where there is almost none today: by the 2090s under SSP2-4.5, most people in Côte d'Ivoire, Cuba, the Dominican Republic, Paraguay, South Sudan, Gabon and Congo live with at least one hazard.

High-income regions are not spared. North America's exposed population nearly quadruples under SSP2-4.5, to 0.16 billion, as humid heat spreads across the southern United States. Europe and Central Asia remain low on these thresholds, which capture humid heat, heavy rain and long dry spells but not Europe's dry-heat extremes, so this understates its risk.

Older people are more vulnerable to heat: their bodies regulate temperature less well, more of them live with heart, lung and kidney disease, and many live alone. Young children are more vulnerable to floods, drought, malnutrition and infection. Across the century, the balance between these two groups changes completely. In 2020, 171 million children under 5 and 136 million people aged 65 and over lived with frequent humid heat. By 2030, older people outnumber children in every pathway, and by the 2090s, 1.3 billion people aged 65 and over live with frequent heat under SSP2-4.5 against 0.25 billion young children: more than five to one.

The shift is driven by Asia's ageing. India's heat-exposed population aged 65 and over rises from 63 million to 450 million under SSP2-4.5, as its over-65 share climbs from 7% to 29%; China's from 36 million to 250 million. Bangladesh, Pakistan, Nigeria, Indonesia, the Philippines, the United States and Viet Nam follow. Under SSP5-8.5, with fast ageing and the most heat, 2.0 billion people aged 65 and over live with frequent heat and 1.5 billion with dangerous heat. This ageing of exposure is the century's least discussed pattern: the countries that do best on fertility and longevity will have the oldest exposed populations.

Children's exposure follows a different path, set by fertility. Under SSP1-2.6, with fast-falling birth rates, the number of young children living with frequent heat halves to 92 million. Under SSP3-7.0, with high fertility in hot countries, it rises to 590 million: 167 million in India, 76 million in Nigeria, 40 million in Pakistan. These children would be growing up where heat already limits outdoor play, schooling and work, and where malnutrition and diarrhoeal disease rise with temperature. The UN projects that people aged 65 and over will outnumber children under 18 by the late 2070s.

Protection takes many forms: early warnings, safer housing, cooling, health care, social protection, functioning government. Few can be projected country by country for a century. Education can: the Wittgenstein Centre projects the schooling of every age cohort under each SSP, and adult schooling is closely tied to many of the others. Educated households heed warnings, prepare, recover faster and demand better services; educated societies build stronger institutions. Lutz and colleagues argued in 2014 that universal education is key to climate adaptation, and this platform's own analysis of disaster deaths found that schooling, not income, carried the protective effect within countries.

This platform's survivability report estimated the relationship within countries: since 1960, each extra year of average adult schooling has gone with 32% fewer disaster deaths per thousand people affected (elasticity −0.38). Applied here, a country's protection factor in each future year is exp(−0.38 × the gain in schooling since 2020). The SSPs differ widely: average adult schooling rises from 8.1 years to 13.3 under SSP1, 12.5 under SSP2 and 13.4 under SSP5, but only to 10.0 under SSP3. In low-income countries the gap is wider still: 10.7 years under SSP2 against 7.0 under SSP3.

The estimate is strong but uncertain (t = −2.4), and it comes from disaster mortality, not heat or chronic exposure. Schooling also saturates: in high-income countries it is already 12–13 years and can rise little, so protection there must come from elsewhere, from cooling, health systems and planning. The simulator in section 8 lets readers weaken or strengthen the elasticity, and section 5 shows how much the conclusions depend on it.

The change in harm can be split into four drivers. Population size is how much exposure would grow if everyone kept their current location and climate; redistribution is the extra from population growing faster in hazardous places; climate is the extra from hazards spreading; protection is the reduction from rising schooling. Under SSP2-4.5, exposure to at least one hazard rises 82% (0.60 log points): climate contributes 61% of that, population size 32% and redistribution 7%. For heat alone, climate contributes 79%. Protection, at −1.85 log points, more than offsets all three, cutting harm by 71%.

SSP3-7.0 is different. Exposure rises 170%, with population size (47%) as large a driver as climate (42%) and redistribution adding 11%, because population grows fastest in the hottest places. Protection grows least. The two nearly cancel: harm falls 7% overall, but turns up after 2050, and heat-related harm rises 39%. In SSP1-2.6 and SSP5-8.5, slower population growth and fast education cut harm by 84% and 71%.

These conclusions rest on protection. With an elasticity half as strong (−0.2), harm falls under SSP1, SSP2 and SSP5 but rises 51% under SSP3. With a weak elasticity (−0.07), it rises in every pathway but SSP1: by 24% under SSP2 and 119% under SSP3. Without protection, it rises 26% to 170%. The lesson is not that education will save the century, but that the protective gains the SSPs assume must actually be delivered, and must work as well as they have in the past.

Each country follows one of six patterns, based on its exposure today and the change in its protection-adjusted harm to the 2090s. Protection wins: harm falls by more than 10% because schooling outpaces exposure; under SSP2-4.5, 87 countries home to 5.8 billion people, including India, China, Pakistan, Bangladesh, Indonesia and Egypt. Climate-led rise: harm rises mainly because hazards spread; 10 countries, including the United States, DR Congo, Kenya, Venezuela, Malaysia and Ghana. Growth-led rise: harm rises mainly because population grows; Canada under SSP2, 26 countries under SSP3.

Holding: harm roughly flat, as in Nigeria, Australia, Benin and Israel under SSP2. Emerging exposure: almost no exposure today but most people exposed by the 2090s; 11 countries under SSP2, among them Côte d'Ivoire, Cuba, the Dominican Republic, Paraguay, South Sudan and Gabon, which have little experience of the hazards they will face. Low exposure: little exposure on these thresholds now or later, mostly in Europe and temperate regions (45 countries).

Under SSP3-7.0, the map changes. Only 42 countries (3.9 billion people) are protected; harm rises in 52 countries home to 2.3 billion people today, among them the United States, Indonesia, Brazil, Nigeria, Mexico, the Philippines, Ethiopia and DR Congo. In 63 countries with 2.7 billion people, the schooling gained falls short of what would offset exposure growth, by up to 10 years in South Sudan and 5 or more in South Africa, Ghana, DR Congo and Malaysia. Even under SSP2-4.5, 19 countries fall short, including the United States, where schooling is already near its ceiling.

The first two layers show the share of each country's people living with at least one hazard in the recent climate and in the 2090s under SSP2-4.5. The third shows how many people aged 65 and over will live with frequent humid heat in the 2090s, the ageing of exposure.

The schooling layer shows the projected gain in average adult schooling to 2090 under SSP2. The margin layer shows the protection margin under SSP3-7.0: negative where schooling falls short of offsetting exposure growth. The pattern layer shows each country's pattern under SSP2-4.5.

Countries are blank where data are missing or population is under 1 million. Hover over or tap a country for its card, with all four pathways; clicking loads it into the simulator. The sortable table lists all 158 countries.

The simulator rebuilds the harm index for any country or the world. Exposure comes from the gridded projections for the chosen pathway and hazard. Vulnerability weights let older people and young children count more than working-age adults; at weights of 1, every exposed person counts the same. Protection scales harm by exp(elasticity × the gain in adult schooling since 2020).

The chart shows three paths relative to 2020 (= 100): exposure alone, exposure weighted for age, and the full harm index with protection. The gap between the second and third lines is what protection buys; the gap between the first and second is what ageing (or youth) adds. For India under SSP2, for example, a weight of 3 on people aged 65 and over raises the age-weighted path by about 40%, because India's over-65 share more than quadruples.

Outputs give the exposed population in the 2090s, the change in harm, the share of exposed people who are 65 and over, and the years of schooling needed to offset exposure growth against those the pathway delivers. These are scenario calculations, not forecasts: the elasticity comes from disaster mortality and may differ for heat, and schooling stands in for a wider set of protections.

Deliver the education transition. The difference between SSP2 and SSP3 is largely a difference in schooling, and it decides whether harm falls by 71% or barely moves. Getting every child into school and keeping them there through secondary level, especially in sub-Saharan Africa and South Asia, is adaptation policy, with payoffs that compound over a lifetime and across generations. The countries with the largest protection shortfalls under SSP3, from South Sudan and DR Congo to Mozambique and Somalia, are those where school systems are weakest today.

Protect an ageing exposed population. By the 2030s, older people outnumber young children among the heat-exposed, and by the 2090s more than a billion people aged 65 and over will live with frequent heat. Heat-health action plans, early warnings that reach people living alone, cooling centres, affordable cooling, home visits and primary care for chronic disease must be built now in India, China and the rest of Asia, while populations are still young enough to plan for it. Where schooling has already saturated, as in the United States, these services are the protection that remains.

Prepare the newly exposed, and limit the hazard. Countries facing hazards they have rarely known, from Côte d'Ivoire and Paraguay to Cuba, need warning systems, building codes and health-system plans before the hazards arrive. Children growing up in the hottest, fastest-growing places need nutrition, water, sanitation and schools built for heat. And because climate is the largest single driver of exposure growth, and the gap between SSP1-2.6 and SSP3-7.0 is five billion exposed people, cutting emissions is what keeps the hazard side of the race within the reach of protection.

Hazards: World Bank CCKP CMIP6 multi-model median climatologies, 0.25°, 1995–2014 and 2020–39, 2040–59, 2060–79, 2080–99 under SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5. Frequent humid heat = heat index above 35 °C on ≥ 30 days (hi35); dangerous heat = above 41 °C on ≥ 30 days (hi41); extreme rain = maximum 5-day rainfall ≥ 150 mm (rx5day); long dry spells = ≥ 90 consecutive dry days (cdd). "Any" counts the three distinct hazards (dangerous heat is a subset of humid heat). Population: SSP-consistent gridded projections (Jones and O'Neill, on the CCKP grid), 2020 and the middle year of each period. Countries assigned with rasterised world-atlas boundaries; 158 countries with at least 1 million people.

Age and protection: Wittgenstein Centre v3.3 SSP projections (KC and Lutz), applied at national level: under-5 and 65+ shares and population-weighted mean years of schooling of adults 25+. Protection factor = exp(−0.38 × Δ schooling since 2020), from the within-country elasticity of disaster deaths per 1,000 affected in report 2026-46. Decomposition: exposure under (future climate, future population), (recent climate, future population) and (future climate, 2020 population); Shapley split of climate and population, with population split into size and redistribution; protection term = log of protection-adjusted over unadjusted exposure. Protection margin = schooling gained − ln(exposed people ratio) / 0.38.

Caveats: climatologies are model medians and thresholds are judgements; humid-heat thresholds miss dry heat, so Europe and the Mediterranean are understated; age shares are national, not local; the pyramid uses Wittgenstein population totals, which differ slightly from the gridded totals; the protection elasticity comes from disaster mortality and is uncertain; schooling stands in for a wider set of protections and saturates in rich countries; patterns use thresholds of ±10% and 5%/15% exposure. Exposure is not impact, and the harm index is a relative measure, not a death toll.