n.V1-7.05 | UNIVERSAL SOCIAL PROTECTION FLOORS

01| The largest anti-poverty solution in the world is not a programme. It is the ordinary machinery of a state paying pensions, child benefits, unemployment insurance and health cover to its own citizens out of its own budget, forever, without a theory of change or an evaluation attached. It is worth stating the size of it before anything else in this book part is discussed, because everything else is a rounding error against it.

02| In 2023, countries allocated on average 12.9 percent of GDP to social protection excluding healthcare, plus a further 6.5 percent to healthcare — 19.3 percent of GDP in total. For the first time, more than half the world's population, 52.4 percent, has some form of social protection coverage, up from 42.8 percent in 2015. That is a genuine achievement and it is almost never reported as one.

03| Now the distribution. High-income countries average 16.2 percent of GDP on social protection. Low-income countries average 0.8 percent. Coverage in low-income countries has not risen since 2017, stagnating near 9.7 percent. Some 3.8 billion people have no coverage at all.

04| Set against that, the narrower category of social safety nets — targeted transfers of the kind the previous chapters described — runs at roughly 1.5 to 1.9 percent of GDP across developing and transition countries, with about 2.5 billion people on beneficiary rolls as of 2018.

05| These figures settle an argument that gets made constantly and loosely: that the world does not have a resource problem, it has an allocation problem. For rich and middle-income countries there is a real argument to be had there, because the budgets are large and the coverage is broad, and how the money is split is a genuine choice. For the countries where poverty is worst, the claim is simply false.

06| The International Labour Organization puts the additional cost of guaranteeing basic social protection in low-income countries at $308.5 billion a year — 52.3 percent of their combined GDP. That single figure exceeds total official development assistance from all DAC donors, which was $212.1 billion in 2024, by roughly 45 percent. There is no reallocation of existing spending that closes a gap of that size. In the places where nine people in ten have no coverage, there is nothing to reallocate.

07| This is where most portfolio arguments about anti-poverty spending quietly break. They treat global spending as one pool and ask how it should be divided. It is not one pool. It is two problems that happen to share a vocabulary: in one, coverage is broad and the question is whether the money is well spent; in the other, there is no coverage and the question is where any money at all is going to come from. A recommendation that makes sense in the first is irrelevant in the second.

08| What a floor does, where it exists, is different in kind from what a project does. It is unconditional in the sense that matters — it does not end. No handover, no exit, no evaporation of impact six months after the funder leaves, because there is no funder to leave. Every complaint made later in this book part about programmes that stop is a complaint about the absence of this.

09| What it does not do is target. Floors are expensive precisely because they reach people who could manage without them, and the political economy that makes them durable — broad coverage creating broad constituency, as the previous chapter argued — is the same property that makes them inefficient by the sector's usual metrics. Cost-effectiveness analysis of the kind discussed at the end of this book part will always score a universal instrument badly, and will always be measuring the wrong thing when it does.

10| The open question is not whether floors work. It is why the countries with 0.8 percent are stuck there, and whether that is a fiscal ceiling, a political choice, or an artefact of an aid architecture that funds projects because projects can be attributed to donors and budget support cannot.