India's rate falls every year since 2009. South Africa's low point was 2021. It misses the line in 2028.

Satoria's forecast

Satoria forecast for South Africa: children who die before their fifth birthday, per 1,000 born 32.6 per 1,000 births in 2034, down from 34.7 in 2024 - could go either way. 80% range 23.3 to 45.3.

Two lines cross the target in 2028. One rises first.

On our central path India reaches 24.3 per 1,000 live births in 2028 and the Philippines 24.5, both under the line of 25. Indonesia keeps falling to 17.1 by 2034. South Africa rises to 34.9 in 2026 before turning down to 32.6. The ranges are wide.

Same starting line in 2009. One country kept going.

The rate counts deaths before the fifth birthday per 1,000 live births. It is not a percentage. India was at 64.6 in 2009 and 27.7 in 2024, down in every single year. South Africa started at 64.2, reached 34.1 in 2021, and has risen in each of the three years since.

Child deaths have more than halved. The falling has got slower.

An estimated 4.9 million children died before their fifth birthday in 2024. Deaths before age five have fallen by more than half since 2000. But since 2015 that fall has slowed by more than 60 per cent. The United Nations line is 25 per 1,000 live births by 2030.

Where does South Africa land in 2034?

South Africa cut child deaths from 64.2 per 1,000 live births in 2009 to 34.1 in 2021, then drifted back up to 34.7. Move the slider to where you think our model puts it ten years on, then see the answer and how wide the range around it is.

South Africa has the widest range of the four.

Our published range for South Africa in 2034 runs from 23.3 to 45.3 per 1,000 live births. The bottom is under the line and the top is worse than today. A model that learns the recent pace cannot see a change of policy before the data shows it.

2028 is the year to check.

If India and the Philippines are under 25 per 1,000 live births in the 2028 figures, our path held. If South Africa breaks clearly below 30 before then, we were too cautious, and we will say so in our public record.

What the world measured

An estimated 4.9 million children died before their fifth birthday in 2024. Deaths before age five have fallen by more than half since 2000, which is one of the largest improvements in human health anybody has measured.

The second half of that sentence is the one that gets less attention. Since 2015 the pace of the fall has slowed by more than 60 per cent, and in 2024 sub-Saharan Africa accounted for 58 per cent of all deaths before age five.

The United Nations target is a single number a country can check itself against: deaths before age five down to 25 per 1,000 live births by 2030.

India and South Africa started level

In 2009 India recorded 64.597 deaths before age five per 1,000 live births and South Africa 64.178. For a comparison across two very different countries that is as close as this data gets.

India then fell in every measured year: 46.263 in 2015, 27.746 in 2024. South Africa fell faster at first, reached 34.072 in 2021, and has come in higher in each of the three years since, at 34.136, 34.366 and 34.651. Three years is a short run and it may yet be noise, but it is the only reversal among these four.

Indonesia is the country that has already finished this argument. It was at 36.424 per 1,000 live births in 2009 and 20.624 in 2024, so it went under the United Nations line without needing a forecast.

What our model says next

On Satoria's central path India reaches 24.2945 deaths before age five per 1,000 live births in 2028 and the Philippines 24.4528 in the same year. Both are then under the line of 25, two years before the target date.

India keeps going after that, to 20.5989 by 2034. Indonesia reaches 17.1017. Neither of those is a surprise: both lines have been falling steadily for the whole measured period.

South Africa is the one that does not arrive. Our central path has it rising for two more years, to 34.8962 in 2026, then turning down to 32.5749 by 2034: about two points below today after ten years. The model reads the recent three years rather than the average of the last fifteen, which is why it is cautious here.

How much to trust this

Satoria's yearly model for this measure is built on the Gapminder child mortality series. On data held out from fitting and model choice, across 49 country series, its errors were 71% smaller than assuming no change.

That is a result across many countries, not a promise about one. The published range for South Africa in 2034 runs from 23.3251 to 45.303 per 1,000 live births, and every one of the four ranges still contains the country's own 2024 value. The flat South African line is the most likely path, not a finding.

One more thing the number is not: it counts deaths per 1,000 live births, not a percentage, and it says nothing about how many children a country has.