Sustainable Development as Global Risk Reduction

Between 2000 and 2010 the HDI of the average person rose by about 7.5 thousandths a year, the fastest pace on record, as child deaths fell, schooling spread and incomes rose across Asia and Africa. The pace slipped to 5.5 a year in 2010–2019 and to 3.6 in 2019–2023, after the pandemic, the cost-of-living crisis, debt distress and a run of conflicts. UNDP's 2025 Human Development Report describes the slowdown as the sharpest outside crisis years since its records began in 1990. Its projections for 2024 show stalled progress in every region.

The slowdown is steepest at the bottom. In the 26 low-HDI countries, home to 910 million people, annual gains fell from 7.8 thousandths in the 2000s to 4.5 in the 2010s and 2.1 in 2019–2023. Madagascar and Yemen went backwards over 2015–2023, Afghanistan stood still, and Benin, Mali and Burundi barely moved. Twenty-two countries had a lower HDI in 2023 than in 2019, among them Palestine, Lebanon, Libya, Venezuela, South Sudan, Afghanistan, Madagascar, Mali, Sudan, Haiti and Ukraine, most of them hit by conflict or economic collapse.

As a result, the gap between the average person in low- and very-high-HDI countries stopped closing. Among the 146 countries with data for every year since 2000, it narrowed from 0.453 in 2000 to 0.406 by 2016 and then stalled. After the pandemic year, when very-high-HDI countries also fell back, it edged up from 0.398 in 2020 to 0.401 in 2023. UNDP, using its own group averages, reports four consecutive years of widening. Medium-HDI countries kept a relatively fast pace (6.5 thousandths a year in 2019–2023), thanks largely to India and Bangladesh. The countries that sustained progress, such as Bangladesh, Rwanda, Côte d'Ivoire, India, Morocco and Viet Nam, show that the slowdown is not inevitable.

Exposure to climate hazards depends on geography, but whether a hazard becomes a disaster depends on people's capacities. Between 2000 and 2024, a median of 2.3% of people in low-HDI countries were affected by a climate disaster each year, against 1.6% in medium-, 1.3% in high- and 0.04% in very-high-HDI countries. Across countries, each 0.1 higher HDI goes with about 63% fewer people affected. Deaths from floods, storms, droughts, landslides and wildfires fall too: a median of 1.6 per million a year in low-HDI countries and 0.2 in very-high-HDI ones, about 38% less per 0.1 HDI.

Cross-country comparisons mix development with geography, so the within-country test matters more. Following each country through seven five-year periods since 1990, and removing both fixed national characteristics and global trends, a 0.1 rise in HDI is associated with about 40% fewer people affected by climate disasters, a statistically significant estimate. It is also associated with about 23% fewer deaths, an estimate too imprecise to be distinguished from zero. Recorded disaster deaths are dominated by a few catastrophic events, which makes them noisy.

The mechanisms are well documented. Lutz, Muttarak and Striessnig found that education, especially women's, predicts lower disaster mortality more strongly than income. Better-educated people understand warnings, prepare and evacuate, and have more ways to recover. Health systems treat the injured and prevent outbreaks after floods. Higher incomes pay for sturdier houses and savings. Heat deaths are excluded here because rich countries record them far more completely than poor ones. Including them erases the apparent gradient, which is a reminder of how incomplete the global disaster record is.

Health vulnerability falls steeply with human development. Under-five mortality has a median of 65 per 1,000 live births in low-HDI countries and 5 in very-high-HDI ones. That gradient is partly built into the index, since child deaths lower life expectancy. Independent measures show the same pattern. Communicable, maternal, neonatal and nutritional conditions account for a median of 52% of deaths in low-HDI countries and 17% in very-high-HDI countries, about 25% less per 0.1 HDI. Out-of-pocket spending, which pushes families into poverty when someone falls ill, accounts for 46% of health spending in low-HDI countries and 21% in very-high-HDI ones. In 2023, 4.8 million children still died before their fifth birthday.

The pandemic broke the pattern. Life expectancy fell by a median of 2.6 years between 2019 and its 2020–2021 low in high-HDI countries, and 1.7 years in very-high-HDI ones, against 0.9 in medium- and 0.5 in low-HDI countries. The worst falls were in Bolivia, Paraguay, Ecuador, Mexico and Peru. WHO estimated 14.9 million excess deaths in 2020–2021, 81% of them in middle-income countries and 4% in low-income ones. Older populations, dense cities, informal work that made distancing impossible and patchy health systems all played a part. So, in part, did weaker death registration in the poorest countries, which may understate their losses.

The lesson is that human development protects against chronic and familiar risks, such as infectious disease, malnutrition and maternal death, but not automatically against novel ones. Pandemic resilience depends on specific capacities: surveillance, laboratory and vaccine capacity, surge capacity in hospitals, trusted public-health communication, and income support so people can afford to stay home. Middle-income countries with high HDI had built some of these and not others.

The people most vulnerable to economic shocks are not only the extremely poor but those just above the line, who can fall below it after one failed harvest, illness or job loss. In low-HDI countries a median of 51% of people live on between $3.00 and $8.30 a day, and in medium-HDI countries 46%, against 23% in high- and under 1% in very-high-HDI countries. Each 0.1 higher HDI goes with 62% fewer near-poor people. In population terms the band is enormous: more than 1.6 billion people in medium-HDI countries alone, many of them in India, Bangladesh and Nigeria.

Social protection is the buffer that keeps the near-poor from falling. Coverage, measured as the share of people receiving at least one benefit in World Bank household surveys, has a median of 16% in low-HDI countries and 28% in medium-HDI countries, against 60% in high- and 71% in very-high-HDI countries. The ILO estimates that 52.4% of the world's population had at least one form of social protection in 2023, the first time above half, leaving 3.8 billion people with none. In low-income countries coverage has stagnated at about 10% since 2017.

Human development does not, however, make economies less volatile. The median standard deviation of annual growth in GDP per head over 2000–2024 is between 3.1 and 3.3 points in every HDI tier, and the share of years with contraction shows no clear gradient either. Volatility comes from commodity dependence, conflict, financial crises and policy. Countries need countercyclical fiscal space, diversified economies and pre-arranged crisis finance in addition to human development. Development makes the same shock less dangerous for households, but it does not prevent the shock.

Bringing the estimates together shows a clear divide. The vulnerabilities that come from deprivation fall steeply with human development: people affected by climate disasters, the near-poor, child mortality, communicable disease, climate-disaster deaths and out-of-pocket health costs. These are the chronic risks that make every shock worse. For them, human development is a powerful and general-purpose form of risk reduction, and the gradients are large. A 0.1 rise in HDI, roughly the distance between Senegal (0.530) and Kenya (0.628), or between India (0.685) and Sri Lanka (0.776), goes with risk levels a quarter to two-thirds lower.

A second group does not fall, or rises. Growth volatility and the frequency of recessions show no gradient. Life expectancy lost in the pandemic rises with HDI, and recorded heat deaths, not shown here, rise with HDI because of reporting. These are the tail and systemic risks: novel pathogens, financial and commodity shocks, and extreme heat in ageing societies. Human development does not remove them, and they need dedicated institutions.

That divide is the core of the policy case. Accelerating human development is the most effective way to cut chronic vulnerability, and it is also what poorer countries want for its own sake. It is not a substitute for preparedness, risk finance and macroeconomic buffers. The best-protected societies combine both, high human development and specific shock institutions. The within-country estimate for people affected (−40% per 0.1 HDI) is smaller than the cross-country one (−63%), as expected once geography is held fixed, but it points the same way.

A risk-reduction strategy built on development would defeat itself if it raised emissions enough to multiply climate hazards. The data show it need not. Twenty-three of the 61 very-high-HDI countries emit less CO₂ per person than the world average of about 4.7 tonnes. They include Costa Rica (HDI 0.833, 1.6 tonnes), Albania (0.810, 1.8), Uruguay (0.862, 2.3), Georgia, Armenia, Panama, Mauritius, Sweden, Switzerland, Portugal, France and the United Kingdom. Twelve countries combine an HDI of at least 0.7 with emissions below 2 tonnes, among them Sri Lanka, Peru, Paraguay, the Philippines, Morocco and Kyrgyzstan.

The relationship between HDI and emissions is steep at the top and nearly flat at the bottom. The median low-HDI country emits 0.17 tonnes per person, the median medium-HDI country 0.67 and the median high-HDI country 2.6. Most of the gain from low to high human development comes from health, schooling and basic infrastructure, which need relatively little energy. The large emissions come with very high incomes and fossil-heavy energy systems, and those are a choice, not a precondition.

UNDP's planetary-pressures-adjusted HDI makes the same point from the other direction: the median very-high-HDI country loses 0.144 points once its CO₂ emissions and material footprint are counted. For low- and medium-HDI countries, the path that combines rapid human development with clean energy, compact cities and efficient industry is now cheaper than ever. Section 7 shows what accelerating along it would buy.

What would acceleration buy? A simple scenario compares two futures for each country to 2035. In the first, HDI keeps rising at its 2015–2023 pace. In the second, each country returns to the fastest pace it achieved in any of 2000–2010, 2010–2019 or 2015–2023, a pace it has already shown it can sustain. The HDI changes are then translated into vulnerability using the cross-country gradients from section 5, applied to each country's current levels. Population is held at 2023, so the results show changes in rates, not the effect of population growth.

For the 26 low-HDI countries, the average HDI would reach 0.598 by 2035 at best pace, against 0.539 at current pace. The near-poor population would shrink to about 176 million rather than 311 million (from 418 million today). People affected by climate disasters each year would fall to about 9 million rather than 16 million, under-five mortality to about 35 per 1,000 rather than 50, and the communicable-disease share of deaths to 36% rather than 42%. For the 34 medium-HDI countries, with 2.5 billion people, the near-poor population would fall to about 590 million rather than 830 million.

These numbers are illustrative. The gradients are associations, not causal effects, and applying them to future changes assumes vulnerability responds to development as it has varied across countries. Climate change will also raise hazards over the same period. The scenario does show the order of magnitude: hundreds of millions fewer people one shock from poverty, and millions fewer hit by disasters each year, from a pace of progress these countries have already achieved.

The HDI layer shows the familiar map, with low human development concentrated across the Sahel, Central and East Africa, plus Afghanistan, Pakistan, Yemen and Haiti. The pace layer shows where progress has continued since 2015, including South Asia, much of West Africa and parts of the Middle East, and where it has stalled or reversed, as in Madagascar, Yemen, Afghanistan, Lebanon, Venezuela and Myanmar.

The vulnerability layers broadly mirror the HDI map, but with telling exceptions. The people-affected layer highlights South and South-East Asia and the drylands of Africa. The disaster-death layer adds Central America and the Caribbean. The near-poor layer shows how much of South Asia and Africa lives just above extreme poverty. The child-mortality and social-protection layers show the health and safety-net gaps, and the CO₂ layer shows that most of the countries with the most vulnerability emit the least.

Hover over or tap a country for its card: the HDI from 1990 to 2023 as a sparkline, its tier and pace, its climate, health and economic vulnerability, and its emissions. Selecting a country also loads it into the simulator in section 9. The sortable table below lists every country.

Choose a country and a pace of human development. The default is its 2015–2023 pace. The slider multiplies that pace, and the button sets it to the country's best decade since 2000. The simulator projects HDI to the target year in a straight line, capped at 1, and applies the cross-country gradients to the country's current vulnerability. It shows the near-poor share, the share of people affected by climate disasters each year, under-five mortality and the communicable-disease share of deaths, at the current pace and at the chosen pace.

For countries whose HDI is falling, such as Madagascar, Afghanistan or Yemen, the current-pace projection shows vulnerability rising. Restoring even modest progress there reverses the direction. For fast-progressing countries such as Bangladesh, India or Côte d'Ivoire, the simulator shows how much faster still the dividend would accrue. The gradient behind each projection is shown with its confidence interval in section 5.

The simulator cannot show what drives the pace, which is the subject of section 10: education, health systems, social protection, clean infrastructure and peace. It also deliberately omits growth volatility and pandemic losses, which the data show do not fall with HDI and need their own institutions.

The first investments are the components of human development that most directly reduce vulnerability. Universal schooling of good quality, especially for girls, cuts disaster mortality and raises adaptability, and it does so for generations. Primary health care, immunisation, nutrition, and safe water and sanitation cut child mortality and communicable disease, the largest health risks in low-HDI countries. Clean, reliable electricity and clean cooking power clinics, schools and businesses while reducing pollution and emissions. These are also the investments with the largest development returns, so the risk dividend comes at no extra cost.

The second set is the institutions that development does not automatically supply. Social protection that covers the near-poor and scales up when a drought, flood or price shock hits keeps people from losing their assets and their children's schooling. Today it covers only about a sixth of people in low-HDI countries. Pandemic preparedness (surveillance, laboratories, surge capacity and income support) and macroeconomic buffers such as fiscal space, contingent credit and pre-arranged disaster finance address the tail risks that HDI does not reduce. Early warning systems turn forecasts into action.

The third is the enabling environment that sets the pace: peace, debt relief and fiscal space, and finance. Most of the countries whose HDI fell after 2019 were in conflict or economic collapse. UNCTAD estimates that 3.3 billion people live in countries that spend more on debt interest than on health or education. Restoring the pace of the 2000s requires debt solutions, concessional finance at scale and a sustainable energy path that keeps development from adding to climate risk. Human development is the most broadly protective risk-reduction strategy available, and it is under-financed precisely where it would protect the most people.

Coverage: 157 countries with at least one million people and an HDI for 2023. Human development comes from the UNDP HDR 2025 composite indices time series, 1990–2023: HDI, its components, the planetary-pressures-adjusted HDI, production CO₂ per person and material footprint. Tiers are UNDP's 2023 classification. Pace is the average annual HDI change, population-weighted for groups.

Vulnerability measures:

• Climate: EM-DAT (via Our World in Data), 2000–2024. People affected by, and deaths from, droughts, floods, landslides, storms and wildfires (deaths exclude extreme temperature), per year and per head.
• Health: under-five mortality (latest), deaths from communicable, maternal, neonatal and nutritional conditions (% of deaths, latest 2015–21), out-of-pocket share of current health expenditure, and life expectancy in 2019 minus its 2020–21 minimum (all WDI).
• Economic: near-poor share, PIP 2024 headcount at $8.30 minus at $3.00 (2021 PPP). Growth volatility, standard deviation of annual GDP-per-head growth 2000–2024. Social protection coverage, WDI/ASPIRE, latest 2010–23.

Gradients: OLS of log vulnerability on HDI (2023, or 2019 for pandemic losses), reported as % change per 0.1 HDI with 95% intervals. Within-country: five-year periods 1990–2023, log outcomes on period-average HDI with country and period fixed effects and country-clustered errors. Acceleration scenario: HDI 2023 plus 12 years at the 2015–2023 pace or the country's best of 2000–10, 2010–19 and 2015–23, capped at 1. Vulnerability is scaled by exp(gradient × HDI change), with population fixed at 2023.

Caveats: associations, not causal effects. Child mortality overlaps with the HDI's life-expectancy component. Disaster records under-count events and deaths, especially heat deaths in poorer countries. Pandemic life-expectancy estimates for many low-income countries are modelled. Projections ignore climate change and population growth.