Divide one unemployment rate by another and you get a number almost nobody publishes

Every country publishes its youth unemployment rate. Every country publishes its overall one. Almost nobody divides the first by the second.

Do it and you get a single number: how much harder it is to be young in that labour market than to be an adult in the same one. Britain's is 3.09. Germany's is 1.85. Both countries have low adult unemployment. Only one of them passes that on to its young people.

The ratio is our arithmetic, not the International Labour Organization's, and we say so wherever it appears. Everything that follows rests on it, so here is exactly what it is made of.

The series. Two indicators from the World Bank's World Development Indicators: youth unemployment for ages 15 to 24, and total unemployment for ages 15 and over. Both come from the same underlying source, the ILO's Modelled Estimates database, which the World Bank last drew on 17 January 2026. We read that at the source rather than assuming it.

Modelled estimates, then, rather than the figures each country reports itself. That is a deliberate choice and it cuts both ways. National figures come from national surveys with different definitions, different age bands and different frequencies, so a two hundred country comparison built on them would be comparing measuring instruments as much as labour markets. The modelling buys comparability and costs a layer of directness. For a ratio, where both halves have to be measured the same way or the answer is noise, comparability wins.

The check. Before building anything on top, we tested the pull against nine European countries whose 2025 figures we already held by a separate route, the German statistical office's country profiles. All nine reproduce. The largest disagreement on either rate is 0.05 percentage points, which is what rounding to one decimal place produces and nothing more. That tolerance is tight enough to answer a question we could not answer before: those published profiles are drawing on the same modelled series.

One thing did not survive the check. The rounded table had Switzerland at 1.8 and Germany at 1.9. Unrounded, Germany is 1.85 and Switzerland is 1.86, so Germany is the lower of the two. The ordering between them was an artefact of rounding, and without the check we would have repeated it.

What is in the table. 6,531 country-years. 187 countries. 1991 to 2025.

Thirty countries have no rows at all: Andorra, Monaco, San Marino, Liechtenstein, the Isle of Man, Gibraltar, the Faroes, Greenland, Kosovo, and twenty-one small island states and dependencies. None runs a labour force survey the ILO models from. No economy of any size is missing.

Five countries stop before 2025. Ukraine ends in 2021, Sudan and the West Bank and Gaza in 2022, Lebanon and South Sudan in 2023. The modelling stops where the survey stops, and the survey stops where the state does. That is worth saying plainly rather than burying: the places most likely to be failing their young are the places this measure cannot see.