Convergent Development as Polycrisis Risk Reduction

The frontier, the median HDI of today's very-high-HDI countries, rose from 0.793 in 2000 to 0.894 in 2019 and 0.900 in 2023. For the 52 vulnerable countries with a complete series, the median gap to that frontier was 0.354 in 2000, 0.343 in 2010 and 0.323 in 2015, then reopened to 0.331 in 2019 before easing to 0.327 in 2023, partly because the pandemic stalled the frontier itself. For the 19 low-HDI countries the gap fell from 0.434 to 0.393 by 2015 and has widened every period since, to 0.403 in 2023.

Measured as an annual rate, countries that were vulnerable in 2000 closed a median 0.36% of their gap a year in 2000–2010 and just 0.10% a year in 2010–2019. Over 2000–2023 the median was 0.32% a year. At that pace, a gap takes about 200 years to halve. The faster 2019–23 figure (0.55%) mostly reflects the frontier's own pandemic setback, not acceleration among the vulnerable.

Fragile and conflict-affected states fared worst: their median gap was the same in 2023 (0.340) as in 2000 (0.354) after allowing for the frontier's rise, and they show the most reversals. Convergence is not automatic. The 2000s, a decade of commodity booms, debt relief and the Millennium Development Goals, were the exception, not the rule.

A simple test of convergence asks whether countries that start lower gain faster. For each component and period, the chart shows the slope of annual gains on starting levels as a t-statistic. Negative bars mean convergence, positive bars divergence, and bars beyond ±2 are statistically clear. Life expectancy converged in every period, with the strongest effect in 2019–23 as the pandemic hit longer-lived populations hardest. Expected years of schooling converged strongly in the 2000s (t ≈ −7), less strongly in the 2010s and not at all since 2019.

Income is the weak component. GNI per head converged in the 2000s, when poorer economies grew about 0.6 percentage points a year faster per log unit of lower starting income. In the 2010s there was no significant relationship, and in 2019–23 incomes diverged (t ≈ +2.5): the pandemic, the 2022 price shock and the rise in interest rates hit poorer economies' growth harder. Because the HDI averages health, education and income, stalled income convergence drags the index even as health keeps improving.

This pattern defines the task. Health and schooling gains are within reach of public policy and global partnerships, and they have proved resilient. Income convergence depends on growth, which is far more exposed to global shocks, commodity cycles and debt. Convergent development must therefore protect the growth engine from shocks as well as invest in people, which is where risk reduction comes in.

Human development does not only move forward. Between 1991 and 2019, outside the pandemic, countries with an HDI below 0.70 saw it fall in about 9–12% of years. Countries above 0.80 saw a fall in 6% of years. The size of the fall differs more: the average decline below 0.45 was 0.0096 points, nearly four times the 0.0026 above 0.80. Combining frequency and depth, the expected annual HDI loss from reversals is about five times larger in countries below 0.55 than in those above 0.80.

Recovery takes longer at lower levels too. The median reversal in countries between 0.45 and 0.70 took five years or more to regain the previous peak, against one year above 0.80. Conflict drove nearly half the reversals in countries below 0.45, a fifth between 0.45 and 0.55, and almost none above 0.80. Economic crises, epidemics such as HIV and Ebola, disasters and political collapse account for most of the rest.

The pandemic was a reversal for most of the world: 56% of countries had a lower HDI in 2021 than in 2019, including 81% of high-HDI countries and 32% of low-HDI ones. For the 87 countries that were vulnerable in 2000, reversals have cost 1.63 HDI points in total since 2001, against 12.57 points of gross gains, or 13% of all progress. The pandemic accounts for 0.76 of those points, conflict for 0.32 and other shocks for 0.55. Only 12 of the 87 avoided any reversal.

Do reversals actually slow convergence, or are they just noise around a trend? Among the 87 countries vulnerable in 2000, those with no reversal in 2001–2019 closed their gap at a median 0.71% a year over 2000–2023. Those with one reversal managed 0.22%, those with two 0.26%, and those with three or more lost ground (−0.24%). Some of this reflects reverse causation, since weak states both reverse and stagnate. But the gradient is steep, and the 33 reversal-free countries include most of the fastest convergers.

The link runs the other way too. Across 156 countries, HDI in 2023 and INFORM 2026 risk are correlated at −0.84: each 0.1 of HDI goes with about one point less risk on INFORM's ten-point scale. The median risk score is 7.8 below an HDI of 0.45, about 5.7 between 0.45 and 0.70, 4.4 between 0.70 and 0.80, and 2.6 above 0.80. Higher human development comes with more coping capacity, less vulnerability and, as section 3 showed, shallower and shorter reversals.

Together these form a loop. Low human development means frequent, deep reversals. Reversals erase gains and slow convergence, which keeps human development low. The loop also runs the other way: steady progress builds capacity, capacity shortens and softens reversals, and fewer reversals allow faster progress. Convergent development and risk reduction are the same strategy seen from two ends, so they should be planned and financed together.

Among the 87 countries vulnerable in 2000, the fastest convergers closed their gap at 1–4% a year, three to thirteen times the median. Türkiye and China led, followed by Kazakhstan, Bangladesh, Eswatini, Armenia, Angola, Vietnam, Rwanda, Morocco and India. Most share a feature: no reversals, or at most one, between 2001 and 2019. Bangladesh, Vietnam, Morocco and China had none. Rwanda raised its HDI from 0.34 to 0.58 with a single reversal, and India from 0.50 to 0.69 with one.

Eswatini is the exception that proves the rule: it converged quickly from a deep HIV-driven collapse once treatment spread, a recovery from reversal, not an absence of it. At the other end, the slowest convergers are defined by repeated reversals: Syria and Yemen through war, Haiti through disasters, political crisis and violence, and Madagascar through political crises and drought. Each had three to seven reversal years.

The fast convergers followed different models: export manufacturing in Bangladesh and Vietnam, state-led investment in China, health and governance reform in Rwanda, oil in Angola. What they shared was continuity. Steady progress, protected from shocks or quickly restored after them, compounds. That is why risk reduction, through conflict prevention, macroeconomic stability, social protection and disaster preparedness, belongs inside a convergence strategy.

Four scenarios project the 60 vulnerable countries (HDI below 0.70 in 2023, 3.4 billion people) to 2050, with gains proportional to each country's distance from the maximum. Recent pace continues each country's 2015–23 trend. Pre-2015 pace restores its 2000–15 trend. Fast-converger pace lifts each country at least to the gross pace of the top quarter of vulnerable countries in non-reversal years, with reversals at the historical frequency and depth for its HDI level. Convergent with risk reduction adds a halving of both reversal frequency and depth.

At the recent pace, the population-weighted HDI of these countries rises from 0.613 to 0.692 by 2050, and 443 million people still live in countries below 0.55. Restoring the pre-2015 pace gets to 0.717. Fast-converger pace reaches 0.731 and empties the low-HDI group by 2050. Adding risk reduction reaches 0.749, cuts the low-HDI population to 16 million by 2040, and brings 38 of the 60 countries above 0.70 by 2050, against 17 at the recent pace.

Because HDI is so closely tied to INFORM risk, the scenarios also imply a large fall in polycrisis exposure. Using the cross-country relationship, the population of vulnerable countries whose HDI implies high risk (an INFORM score of 5 or more) would grow from 1.7 billion today to 2.3 billion in 2050 at the recent pace, as their populations grow. Under convergent development with risk reduction it would fall to about 0.2 billion. These are associations projected forward, not forecasts, but they show the stakes.

The HDI layer shows the 2023 index. Vulnerable economies, below 0.70, cover most of sub-Saharan Africa, South Asia apart from Sri Lanka, and parts of the Middle East, South-East Asia, Central America and the Caribbean. The convergence layer shows each country's annual gap-closure rate since 2000. Fast convergers stand out in East and South Asia and parts of Africa, while countries that fell further behind cluster in conflict zones.

The reversals layer counts years since 2000 in which HDI fell, and the loss layer shows the share of gross gains erased by them. Both are highest in conflict-affected and politically unstable countries. The risk layer shows INFORM 2026, which mirrors the HDI layer closely. The 0.70 layer shows the year each vulnerable country reaches an HDI of 0.70 under convergent development with risk reduction.

Hover over or tap a country for its card: HDI from 1990 with reversal years marked, its components in 2000 and 2023, convergence and reversal statistics, risk, and the year it reaches 0.70 under each scenario. Clicking a country loads it into the simulator in section 8. The sortable table below lists all 157 countries.

Choose a country. The pace slider sets the share of its remaining distance to an HDI of 1 that it closes each year in non-reversal years, starting from its own 2000–2019 pace; the top quarter of vulnerable countries managed about 1.9%. The reversal slider sets how much the historical frequency and depth of HDI falls for its level are reduced, through conflict prevention, macroeconomic buffers, social protection and disaster preparedness.

The chart shows the expected HDI path to 2050 against the frontier and the 0.70 and 0.80 thresholds, with the recent-pace path for comparison. The outputs show HDI in 2050, the year 0.70 is reached, the expected HDI points lost to reversals along the way, and the INFORM risk score implied by the 2050 HDI under the cross-country relationship.

Expected paths average over reversals, so real paths will be bumpier. Try Nigeria or Pakistan, where the recent pace leaves them below 0.70 in 2050, against Bangladesh, which crosses it within a few years. Try a fragile state such as Yemen or Haiti to see how much reversal reduction alone is worth.

Protect the gains. Reversals erased 13% of vulnerable countries' progress since 2000, and avoiding them was the clearest marker of fast convergence. Conflict prevention and early peacebuilding, social protection that scales up automatically in a shock, health systems that keep services running in emergencies, disaster early warning, and macroeconomic buffers each reduce the frequency or depth of reversals. Their returns should be counted in convergence years saved, not only in losses avoided.

Restart income convergence. Health and schooling have kept converging, but income has diverged since 2019, and the HDI will not converge without it. That means investment-friendly macroeconomic stability, debt relief and lower borrowing costs, reliable energy and infrastructure, and trade access. It also means protecting schooling, whose convergence has stopped since the pandemic, with catch-up learning and attendance programmes.

Plan and finance them together. Convergent development and risk reduction are one loop, so they belong in one plan and one financing framework: contingent finance and climate-resilient debt clauses that keep investment going after shocks, budgets that protect health and education spending in downturns, and development partners who measure success by gaps closed and reversals avoided. The fast convergers show it can be done within a generation.

Data: UNDP HDR 2025 composite-indices time series, 1990–2023 (HDI, life expectancy, expected and mean years of schooling, GNI per head in 2021 PPP dollars), for 157 countries with at least one million people. Frontier: median HDI of the 61 countries classed as very high in 2023 with complete 2000–2023 data. Vulnerable: HDI below 0.70 in 2023 (60 countries) or, for retrospective analysis, in 2000 (87 countries with complete data). Gap-closure rate = annual percentage change in the log gap to the frontier. Convergence tests regress annual gains on initial levels (income in logs) for 1990–2000, 2000–10, 2010–19 and 2019–23.

Reversal = any annual HDI decline. Probability and depth are computed by HDI band at the start of the year for 1991–2019; recovery = years until the pre-decline peak is regained. Cause: pandemic for 2020–21; conflict if UCDP GED records at least 1,000 deaths in organised violence in the year or the previous year; otherwise other. Fragile = World Bank FY2026 list of conflict-affected countries. Risk: INFORM Risk Index 2026; linear fit of INFORM on HDI 2023 (n = 156, R² = 0.71).

Scenarios: ΔHDI = r × (1 − HDI), with r from each country's 2015–23 net pace (recent), 2000–15 net pace (pre-2015), or the larger of its 2000–19 non-reversal pace and the vulnerable top-quartile pace, minus expected reversal loss by band (fast-converger; halved for the risk-reduction case). Populations from UN WPP 2024 (medium). Caveats: HDI is modelled and interpolated, so annual changes are noisy; reversal counts include data revisions; associations are not causal; scenarios project cross-country relationships forward.