Skip to content
BabZıtunaThe Market Desk

The record · 27 August 2026

Youth unemployment looks stable. Nine million more young people are neither working nor studying.

A rate that moves a tenth of a point is the kind of number that gets called stable and then gets forgotten. But the report carrying it also counts nine million more young people who are neither in work nor in study, and puts youth unemployment at 3.4 times the adult rate - because the adult rate fell while the youth rate rose. Three measures, one generation, and the disagreement is not error but arithmetic: each counts a different denominator. Every figure below is a published value from the ILO report linked at the foot of this piece, read in the report itself.

12.4%the global youth unemployment rate in 2025, up from 12.3% in 2023 - and projected to stay between 12.3% and 12.4% through 2027
257myoung people not in employment, education or training in 2025 - nine million more than in 2023, a share of 20% against 19.7%
3.4the ratio of the youth unemployment rate to the adult rate in 2025, and rising: youth went from 12.3% to 12.4% while adults went from 3.7% to 3.6%

A tenth of a point, and nine million people

In 2023 the global youth unemployment rate reached its lowest level in more than two decades. The ILO's new report shows it has stopped falling: 12.3% in 2023, 12.4% in 2025, and a projection that it stays between those two figures through 2027. Read that line alone and the story is that nothing is happening.

The same report counts the young people who are in neither a job nor education nor training. That group grew by nine million between 2023 and 2025, reaching 257 million - a share of the world's young people that went from 19.7% to 20%. Read that line alone and the story is a crisis.

Both lines are correct, and the reason is in the denominators. The unemployment rate divides by the young labour force: those with a job and those actively looking for one. The NEET share divides by the young population: everyone, including those who stopped looking. A young person who gives up the search leaves the first measure entirely and stays in the second.

Global rates, per cent, from the ILO's Global Employment Trends for Youth 2026. Youth means 15-24; adults means 25 and over. The first two rows are shares of the LABOUR FORCE - people working or actively looking. The third is a share of the whole youth POPULATION, which is why it can rise while the unemployment rate holds. The rows are not three attempts at one number.
What is counted, and out of what20232025
Unemployed aged 15-24, as a share of the youth labour force12.312.4
Unemployed aged 25+, as a share of the adult labour force3.73.6
Aged 15-24 in neither work nor education or training, as a share of all young people19.720.0

Why a steady rate can hide a rising number

The ILO makes this point about its own indicators. The unemployment count covers young people who are out of work, not in education, and looking. NEET adds a much larger group: those with no job and no formal learning who are also not actively searching - the discouraged, and those held back by obstacles such as care responsibilities.

The consequence is counter-intuitive and worth holding on to: more schooling can push the unemployment rate up. When young people move from unemployment into education they leave the unemployed pool without entering employment, which shrinks the labour force the rate is divided by. By the ILO's reading, a falling NEET share is the more reliable good news, because it can only come from more young people working or more of them learning.

This is why "the youth unemployment rate is stable" is a defensible sentence and a poor summary. It is a statement about the young people who are still in the market. It says nothing about the ones who left it.

Three point four

The clearest signal in the report is not either rate on its own but the distance between them. Between 2023 and 2025 the youth rate rose from 12.3% to 12.4% while the adult rate, for people aged 25 and over, fell from 3.7% to 3.6%. The two moved in opposite directions.

That leaves the youth unemployment rate at 3.4 times the adult rate in 2025, and the report records this ratio rising in Eastern Asia, Northern Africa, Northern America, Northern, Southern and Western Europe, and Southern Asia. The labour market did not get worse for everyone. It got worse for the people trying to enter it, while it improved for the people already inside.

The regional spread is wide. In 2025 youth unemployment stood at 26.2% in the Arab States and 22.6% in Northern Africa; at 15% across Northern, Southern and Western Europe; and at 9.8% in Northern America, up from 8.3% in 2023.

The gap that schooling does not explain

One figure in the report resists the usual explanation. In 2025 the employment-to-population ratio for young men was 43.3%; for young women it was 30.6% - a difference of nearly 13 percentage points.

The report is explicit that this is not a story about access to school: young men and young women are now on par in enrolment. Whatever produces a thirteen-point gap in who holds a job, it is not who was allowed to study. And because so much of that gap sits with young women who are not looking for work rather than looking and failing, it is largely invisible to the unemployment rate and visible in the NEET share.

If you are hiring

Two practical things follow, and neither requires believing any forecast.

The first is that a national youth unemployment rate is a poor read on the pool you can actually reach. It describes people already searching. If you want to know how many young people in a market could work and currently do not, the NEET share is the measure that includes them - and it is the one that moved.

The second is that the report names an erosion of medium-skilled jobs - the roles that have long been the first rung after secondary school or technical training. If that rung is thinning, the entry-level opening you post is doing more work than it used to, and the candidate who applies to it has fewer alternatives than the same candidate had three years ago.

We match people to work, and a measure that ignores the ones who stopped looking would flatter us. How we test our matching for bias is public, and it is the same discipline as the one above: name the ruler before you quote the number.

Earlier editions