Social Safety Nets
Roughly 3.8 billion people — nearly half the world's population — are not covered by any form of social protection, according to the ILO. In 2023, 52.4% of people received at least one benefit, the first time coverage passed half the world. It reaches about 86% in high-income countries and just 9.7% in low-income countries. Public social spending averages about 21% of GDP in the OECD and far less in the poorest countries. The combination of population aging and emerging-market formalization will be the central social-protection challenge of the coming decades.
Key insights
Coverage tracks income — but not perfectly
Low-income countries: 9.7% of the population covered. Lower-middle-income: 32.4%. Upper-middle-income: 71.2%. High-income: about 86%. Within income groups, coverage varies — Bolivia (87%) covers more than Mexico (63%); Mongolia (100%) covers more than China (76%). Political choice and program design matter alongside per-capita income. ILO Recommendation 202 calls for universal social protection floors but only a minority of countries have legislated them.
Aging shifts spending mix toward elderly
OECD social protection spending is increasingly concentrated on pensions and healthcare for elderly — 60-70% of social spending in many advanced economies. Child benefits, family support, working-age unemployment have shrinking shares. This raises questions about generational equity: working-age taxpayers fund elderly transfers, but their own children's social protection has been eroded. The 'generational compact' debate has intensified.
Informal-sector exclusion is the binding constraint in LMICs
Most social-insurance schemes (contributory pensions, unemployment insurance) are designed for stable formal employment. Informal workers (~60% of LMIC employment) are excluded from these schemes — by definition. Non-contributory (tax-funded) programs like cash transfers and old-age pensions can reach informal workers but require fiscal space many LMIC governments lack. Reaching universal social protection requires fundamental redesign, not just expanded coverage of existing programs.
Social protection spending — selected countries (% of GDP)
All forms: pensions, health, unemployment, family, social assistance
Show data table
| % of GDP | |
|---|---|
| France | 31 |
| Italy | 30 |
| Finland | 30 |
| Sweden | 29 |
| Germany | 28 |
| Denmark | 28 |
| Spain | 25 |
| UK | 21 |
| OECD avg | 21 |
| USA | 18 |
| Russia | 15 |
| Brazil | 16 |
| South Africa | 12 |
| World | 13 |
| China | 9 |
| India | 3 |
| Indonesia | 2 |
| Nigeria | 1 |
Key Finding: Nordic countries top the rankings; emerging markets cluster at 5-10%; low-income countries below 5%.
At least one social-protection benefit — coverage rate by income group
% of population covered by at least one benefit, 2023
Show data table
| % coverage | |
|---|---|
| High income | 86 |
| Upper-middle income | 71.2 |
| World | 52.4 |
| Lower-middle income | 32.4 |
| Low income | 9.7 |
Key Finding: Coverage climbs steeply with income, and has barely moved in low-income countries since 2015.
Methodology & caveats
ILO definitions
Social protection floors (ILO Recommendation 202): basic income security across the life course + access to essential healthcare. 'At least one benefit' is the headline coverage measure but says nothing about adequacy. A poorly-funded means-tested program technically counts; so does universal child benefit. Cross-country comparisons should specify both coverage and benefit-level.
Contributory vs non-contributory
Contributory schemes require past contributions (payroll deductions, employer contributions) — most pensions and unemployment insurance. Non-contributory schemes are tax-funded and don't require past contributions — most social assistance, conditional cash transfers, universal old-age pensions. LMICs increasingly mix the two: contributory for the formal sector, non-contributory floors for everyone else.
Measurement issues
Coverage statistics typically count program participants, not beneficiaries served. A family of four with one cash-transfer recipient might count as one beneficiary or four depending on definition. Adequacy isn't measured by coverage at all — Bolivia's nearly universal Renta Dignidad pension pays just ~$50/month, hardly poverty-preventing. WHO/ILO/World Bank are working toward common indicators but harmonization is incomplete.
Why ILO and World Bank coverage figures differ
The ILO counts legal and effective coverage across nine social-security branches, including pensions and contributory schemes, and it counts anyone who receives at least one benefit. The World Bank's ASPIRE database focuses on programmes reaching households in surveys, so it can show who actually received a transfer, including whether the poorest quintile did, but surveys often leave out smaller schemes. The same country can look quite different in the two sources. The largest caveat is adequacy: a tiny transfer counts the same as a full pension. Coverage in low-income countries is also heavily dependent on donor-funded programmes, so aid cuts could reverse gains. The numbers would change most if informal workers were brought into contributory schemes at scale, which is why that is the central policy debate in middle-income countries.
Cite this page
GlobeChart. “Social Safety Nets.” Last updated 2026-10-31. https://globechart.com/poverty/social-safety-nets/
Please also credit the original publisher named under each chart: ILO World Social Protection Report 2024-26. Spotted an error? Tell us.