Polycrisis Resilience: Who Can Afford It?

The median high-income country spends 1.3% of GDP on interest and 5.8% on health, so interest is a fifth of health spending. The median upper-middle-income country spends 1.8% on interest and 3.5% on health, the median lower-middle-income 1.7% against 2.0%, and the median low-income 1.0% against 1.0% (on the latest observation each in 2019–24; the 2022–24 window in section 2, with fewer countries, gives 1.9% against 0.9%). The ratio of interest to health is 0.21, 0.60, 0.95 and 0.71 respectively, and the share of countries above 1 is 4%, 30%, 47% and 31%.

Of the 34 countries, 13 are upper-middle-income, 15 lower-middle-income, four low-income and two high-income (Bahrain and the Bahamas). India (1,438 million people, interest 3.2% of GDP against health 1.3%), Brazil (8.0% against 4.3%), Bangladesh (2.0% against 0.3%), Mexico, the Philippines and Türkiye dominate the people count. The ratios are highest in Sri Lanka (7.1), Malawi (6.8), Bangladesh (6.4), Angola (4.1), Togo, Uganda and Lao PDR (above 3). Countries with big interest bills are often those with thin revenue: the 34 have a median revenue of 17% of GDP, and 23 of the 141 raise under 15%.

The comparison is deliberately blunt. Interest is the price of past borrowing, often to build things; health spending is only one of several resilience investments; and a high ratio can reflect low health spending as well as high interest (Bangladesh spends 0.3% of GDP on health). It is nevertheless the comparison used in international debate (UNCTAD counted about 3.4 billion people in developing countries where interest exceeds health or education spending in its 2025 edition, and later material cites 3.7 billion), and it shows what a fiscal squeeze means in practice: the claim on the budget that cannot be negotiated is larger than the claim that protects people.

In 2010–12 the median low-income country spent 3.3% of its revenue on interest; in 2022–24 it spent 11.2%. As a share of GDP the interest bill rose from 0.5% to 1.9%, while government health spending went from 1.1% to 0.9%. For lower-middle-income countries the interest burden rose from 4.8% of revenue to 8.1% after peaking at 9.4% in 2016–18, and for upper-middle-income countries from 6.5% to 9.0%. In high-income countries it fell from 5.7% to 3.5%. The share of countries with interest above health spending rose from 16% (21 of 133) in 2010–12 to 23% (27 of 116) in 2022–24, after peaking at 24.5% in 2016–18. Windows contain different samples (the latest window has fewer countries because 2024 data are incomplete), so the figures are indicative of direction, not exact paths.

The causes are well known. Many low-income countries borrowed heavily after the global financial crisis, on commercial and bilateral terms, when rates were low; the pandemic and the 2022 food and fuel shocks added debt; and rates rose sharply after 2022 while currencies fell, raising the cost of dollar debt. The World Bank's International Debt Report 2025 finds that low- and middle-income countries paid $741 billion more in principal and interest to external creditors than they received in new money over 2022–2024, the largest net outflow in 50 years, and paid more than $415 billion in interest alone in 2024. UNCTAD finds developing countries' net interest payments at about $921 billion in 2024.

For resilience this matters twice. It raises the cost of finance for the investments that cut future losses, and it makes crisis response harder, because it increases the share of revenue that is already spoken for before the shock arrives. Revenue, not debt, is the denominator: a country raising 13% of GDP in revenue has little room for a 1.9% interest bill. The causal story is one of arithmetic and priority, not a claim that borrowing was unwise: most of it financed investment, and the question is the terms.

Three simple tests. First, did countries with high interest burdens expand public spending less in 2020? Regressing the change in central-government expense (% of GDP) from 2019 to 2020 on the 2019 interest-to-revenue ratio and log income in 129 countries gives −0.08 points of GDP per point of interest burden (standard error 0.03). A country spending 30% of revenue on interest rather than 10% expanded spending by about 1.6 points of GDP less. Income matters more: each doubling of income per person was associated with a 0.9-point larger expansion.

Second, did the heavy-interest countries recover less? Log GDP per person in 2023 relative to 2019 falls by 0.18 percentage points per point of interest burden (standard error 0.08; 135 countries), so a country at 30% rather than 10% was about 3.7% lower in 2023, other things equal. The model explains only 9% of the variation, and the 2019 interest burden is correlated with a country's other vulnerabilities (commodity dependence, currency weakness, conflict) that the regression does not control.

Third, does a rise in interest crowd out health spending? Across 104 countries, the change in health spending from 2019 to 2023 was −0.10 points of GDP per point rise in the interest bill (standard error 0.07): negative, but not distinguishable from zero. Health spending tends to be sticky and low in the poorest countries whatever happens to interest. The squeeze, in other words, operates less as a visible swap of interest for health than as a ceiling on how far governments can expand in a crisis and how much they can invest ahead of one. It is the lack of room that matters for resilience, which is the case for the cost-of-capital and revenue levers in section 7.

Overlaying the INFORM Risk Index (hazard, vulnerability and lack of coping capacity combined) shows that the squeeze and the risk are correlated but not the same thing. Countries with INFORM scores of 5 or more, 36 of the 121 with both data sets, hold 2.9 billion people; 17 of them, with 2.2 billion people, spend more on interest than on health. India alone accounts for 1.4 billion; without it the squeezed high-risk group has about 0.8 billion people, including Bangladesh, the Philippines, Türkiye, Tanzania, Kenya, Myanmar, Uganda, Angola, Ghana, Mozambique and Zambia.

The overlay matters because the countries with the highest risk are also those where an additional shock is costliest and where the capacity to build resilience is thinnest. Some high-risk countries are not squeezed because they raise so little revenue that interest is small beside the need: the Democratic Republic of the Congo, Somalia, the Central African Republic, Burundi, Mali, Burkina Faso and Ethiopia have interest-to-health ratios below 1, and fragile states such as Afghanistan, Sudan, South Sudan and Yemen have no comparable data. For them the issue is revenue, aid and security, not the cost of debt. The squeezed high-risk group is a different case: countries with market access or multilateral borrowing, large populations and growing exposure, where the cost of capital is what limits the investment.

A third group, with high interest but lower measured risk (Brazil, Mexico, Malaysia, Sri Lanka, Senegal), is squeezed while moderately exposed; for them resilience is partly a matter of efficiency and of the composition of spending. The overlay therefore suggests three tracks: debt and cost-of-capital relief for the high-risk, high-interest group; revenue and grant support for low-revenue fragile countries; and spending reform and resilience-linked financing for the middle group.

The calculator is arithmetic on a country's own fiscal position. The programme costs a share of GDP; the calculator converts it to dollars, to a share of government revenue and to a multiple of health spending. It then lets the reader fund it from three sources: relief on the interest bill (the share of interest payments released), additional revenue (as a share of GDP) and grants or concessional finance (a share of the programme). Whatever is left is the unfunded gap, which would have to come from reallocation or new borrowing, and the interest bill is shown as a multiple of the programme for scale.

Benchmarks help calibrate. For the 34 squeezed countries, a programme of 1% of GDP would cost about $117 billion a year, against an interest bill of $495 billion and health spending of $278 billion. A 24% reduction in interest would pay for all of it. For a typical squeezed country, 1% of GDP is about 6% of government revenue. UNEP's Adaptation Gap Report 2025 estimates that developing countries need $310 billion a year for adaptation by 2035 on modelled costs ($365 billion on national plans) against $26 billion of international public flows in 2023, and finds that 59% of those flows were loans.

The sliders are assumptions to be argued about, not forecasts or recommendations. The programme share is a stand-in for what a country would need, which varies; interest relief is not costless to lenders, and not all of it is obtainable; revenue mobilisation takes years and has political limits; and grants are scarce. The calculator shows only the scale of the trade-offs, and where they are tightest: countries with low revenue and a large interest bill have to rely on a mix of all three.

Lower the cost of the capital that is already owed. The evidence in section 2 is that the squeeze is a terms problem as much as a quantity problem. Restructuring and refinancing that lengthen maturities and cut rates, pause clauses that suspend payments after a disaster, and more concessional multilateral lending reduce the interest bill without new borrowing. The arithmetic in section 5 shows the scale: a fifth to a quarter of the squeezed countries' interest bill would fund a programme of 1% of GDP. These countries are not asking for new money so much as for better terms on the old.

Raise and protect revenue. A low-income country raises about 13% of GDP; the squeezed countries 17%. The interest bill is a claim against revenue, so every point of revenue gained makes the burden lighter, and every resilience investment that avoids a loss protects the revenue base. Tax capacity, formalisation and the digital rails of earlier reports are the long-run lever; a ring-fenced share of new revenue for resilience is the political one. Spending reviews that find waste in energy subsidies and untargeted transfers release space too.

Make the money buy resilience, and measure it. Grants and concessional finance are scarce, and 59% of international adaptation finance is lent, so a dollar should go where it is most protective: early warning, social protection that scales in a crisis, health and flood systems. Countries should publish a resilience budget line, with outcome indicators, so that lenders can see what interest relief buys, and relief can be tied to it. The evidence on that link is thin: this dashboard documents the squeeze and its association with weaker crisis response, and cannot show that relief would improve outcomes.

Fiscal data are World Bank World Development Indicators series (retrieved November 2026): GC.XPN.INTP.RV.ZS (central-government interest payments, % of revenue), GC.REV.XGRT.GD.ZS (revenue excluding grants, % of GDP), SH.XPD.GHED.GD.ZS (domestic general-government health expenditure, % of GDP), SE.XPD.TOTL.GD.ZS (government education expenditure, % of GDP), DT.TDS.DECT.EX.ZS (total debt service, % of exports), GC.XPN.TOTL.GD.ZS (central-government expense, % of GDP), NY.GDP.PCAP.KD and NY.GDP.PCAP.PP.KD (GDP per person), NY.GDP.MKTP.CD (GDP) and SP.POP.TOTL (population). Income groups are the World Bank's. For each country the latest observation in 2019–2024 is used (education 2018–2024). Interest as a share of GDP is interest % of revenue times revenue % of GDP. Squeezed means interest above government health spending; the 141 countries with interest, revenue and health data hold 6.4 billion people. INFORM Risk Index, Mid-2026, is the overall risk score (0–10), as used in report 2026-66.

Trends use the latest observation in each of five three-year windows (2010–12 to 2022–24) per country, so the sample varies (133, 142, 143, 140 and 116 countries) and low-income medians rest on 10–14 countries. Tests are OLS cross-sections: change in central-government expense 2019–20 on 2019 interest % of revenue and log GDP per person (PPP, latest); log GDP per person (constant 2015 US$) in 2023 minus 2019 on the same regressors; change in health spending 2019–23 on the change in interest % of GDP and log GDP per person. Standard errors are the classical OLS ones. The calculator is arithmetic: programme = share × GDP; funded = relief × interest bill + revenue gain × GDP + grants × programme; the gap is the unfunded remainder.

Caveats: interest is central-government while health is general-government, so the ratio understates the squeeze in federations; health spending is not the only resilience line; many low-income countries do not report interest and revenue consistently, and the latest year differs by country; 2024 data are incomplete and later restructurings (for example in Sri Lanka and Ghana) are partly in the figures; the tests are associations that do not identify causes, and interest burdens correlate with other vulnerabilities; and the programme sizes in the calculator are assumptions. External benchmarks (UNCTAD, UNEP, World Bank International Debt Report) are quoted from those organisations' releases and differ in definitions from the dashboard's own figures.