Two numbers, one table, opposite signs
Eurostat publishes one monthly unemployment release covering both the European Union and the euro area. Most months the two move together, and a reader can treat either as a summary of "Europe" without doing much damage. June 2026 is not one of those months.
The release reports that, compared with June 2025, youth unemployment increased by 53 thousand in the EU and decreased by 8 thousand in the euro area. That is the wording, and it is the whole story. One aggregate improved. The other, which contains it, got worse.
The levels behind the sentence are in the same table. Unemployed under-25s in the euro area went from 2,359,000 to 2,351,000. Across the Union they went from 2,934,000 to 2,987,000. Subtract one from the other and the population outside the euro area went from about 575,000 to about 636,000: a rise of roughly 61,000. That subtraction is ours, not Eurostat's, and it is the only derived number in this piece.
| Population | Unemployed under-25s | Change on the year |
|---|---|---|
| Euro area | 2,359,0002,351,000 | -8,000 |
| European Union | 2,934,0002,987,000 | +53,000 |
Eurostat publishes no subtotal for the states outside the euro area. This is the only derived number in the piece, and both inputs are rounded to the nearest thousand.
The first objection is Bulgaria, and it does not hold
The euro area is not a loose term for the continent, and it is not the European Union. In June 2026 it is twenty-one of the Union's twenty-seven member states. Bulgaria became the twenty-first on 1 January 2026 - which invites the obvious explanation: a country moved from one side of the line to the other, and the two aggregates diverged because their memberships changed underneath them.
They did not. Eurostat now publishes the euro-area series on the twenty-one-member composition throughout: the euro-area youth figure for June 2025 is 2,359,000 on that basis, in the July release and in the live dataset alike, and the twenty- and nineteen-member series no longer return any data at all. Bulgaria is counted inside the euro area at both ends of the comparison. It cannot be the reason the two lines part.
That leaves six states inside the Union and outside the single currency, and the whole of the divergence sits with them by construction - they are, definitionally, the difference between the two aggregates. What they are not is an explanation, and the table below should be read as a description of who is in that group rather than as an accounting of the 61,000.
| Member state | Under-25 rate | Under-25s unemployed | All ages |
|---|---|---|---|
| Sweden | 24.9% | 177,000 | 8.7% |
| Denmark | 16.6% | 86,000 | 6.5% |
| Hungary | 13.5% | 40,000 | 4.5% |
| Czechia | 12.9% | 44,000 | 3.3% |
| Poland | 12.6% | 141,000 | 3.1% |
| Romania | no monthly figure | 6.3% | |
Objection
Bulgaria adopted the euro on 1 January 2026. Did the two aggregates change composition underneath the comparison?
Verified against the record
No.
Eurostat publishes the euro-area series on the twenty-one-member composition at both ends: 2,359,000 unemployed under-25s in June 2025, identical in the release and in the current dataset. Queries for the twenty- and nineteen-member aggregates return no data at all.
- June 202521 member states
- June 202621 member states
European Union
Euro area
-8,000
The six member states outside the euro area
≈ +61,000 Provenance: Calculation
The euro area is twenty-one of the Union's twenty-seven member states. The remainder is not a published subtotal: it is what is left when one aggregate is subtracted from the other.
What a youth unemployment rate actually divides by
Sweden's 24.9% is the highest under-25 rate published for June 2026. It does not mean that a quarter of Swedes under twenty-five are unemployed, and reading it that way is the most common error made with this statistic.
The denominator is the youth labour force - those working or actively looking for work - not the youth population. Most people under twenty-five in northern Europe are in education and are counted in neither the numerator nor the denominator. A country where students routinely look for part-time work will report a higher youth unemployment rate than one where they do not, without a single additional young person being out of work in any sense a reader would recognise.
Nor is 24.9% "the highest in the Union", though it is the highest number on that page. Romania publishes no monthly figure: the release lists it among five countries reporting quarterly, and its last published under-25 rate, 28.8% in March 2026, was higher than Sweden's. The strongest true statement is the narrower one.
This is also why the level matters more than the rate for the question this piece asks. A count of people can be added and subtracted between aggregates. A rate cannot. The 8,000 and the 53,000 are counts, and that is what makes the comparison legitimate at all.
24.9% does not mean what it looks like
24.9% of all Swedes under twenty-five are unemployed
24.9% of those under twenty-five who are in the labour force are unemployed
- People under 25
- In the labour force
- Working
- Unemployednumerator
- Studying, or otherwise not seeking workoutside the denominator
Counts
-8,000 +53,000
Can be added and subtracted between aggregates. That is what makes the comparison in this piece legitimate.
Rates
24.9% 12.9% 16.6%
Cannot. Each divides by a different labour force, so the difference between two of them measures nothing.
The divergence is solid. The detail underneath it is not.
Two things are true at once here, and reporting only the first would be the easy mistake.
The divergence itself is about as well established as a monthly figure gets. It appears in the release of 30 July and, unchanged, in the Eurostat dataset as it stood on 25 August - two vintages, four weeks apart, both giving minus 8 thousand and plus 53 thousand. Nothing in the intervening revision touched it.
The country detail is a different matter. Czechia's under-25 rate is 12.9% in June against 8.6% in May - a jump of more than four points in a single month, on a base that moved from 28,000 to 44,000 young people. Cyprus's entire count of unemployed under-25s is 2,000. Five member states report quarterly rather than monthly. And Sweden's June rate, printed as 24.9% on 30 July, reads 25.8% in the dataset updated on 25 August - the same month, revised by nearly a point after publication.
A last check makes the limit concrete. The six outside states, as published for June 2025, sum to 584,000 unemployed under-25s; the same group implied by subtracting the two aggregates is 575,000. Nine thousand apart, on figures that are meant to describe the same people. Rounded national counts and a quarterly-reporting Romania are each enough to open a gap that size, and we did not establish which does it here - which is precisely why no share of the 61,000 is assigned to any country in this piece.
What the release does establish is narrower and more useful than any of that: a reader who followed only the euro area figure and a reader who followed only the EU figure learned opposite things about the same year.
Solid
In the release of 30 July and unchanged in the dataset of 25 August. Two vintages, four weeks apart.
- Euro area: 8,000 fewer unemployed under-25s
- European Union: 53,000 more, over the same twelve months
- The euro area is contained in the Union at both ends of the comparison
Less firm
Small national series, one revised after publication, and five member states that report quarterly rather than monthly.
- Czechia: 8.6% in May, 12.9% in June
- Cyprus: 2,000 unemployed under-25s in total
- Sweden: revised from 24.9% to 25.8%
The table in this piece prints 24.9%, because the release is what the piece reports. Neither the minus 8 thousand nor the plus 53 thousand changed between the two vintages.
What the data establishes
- Unemployed under-25s in the euro area fell by 8,000 over the year to June 2026.
- Across the European Union the same count rose by 53,000.
- The euro area is twenty-one of the Union's twenty-seven member states, at both ends of the comparison.
- The difference therefore sits with the six states outside the euro area, and is about 61,000.
What it does not
- Which of the six states accounts for what share of the 61,000.
- That 24.9% is a share of all young Swedes rather than of the youth labour force.
- That the country figures are as firm as the two aggregates: one was revised, and one member state has no monthly figure at all.
Method and what is not claimed
- Two documents, both Eurostat, both linked below: the euro indicators release 3-30072026-BP, "Euro area unemployment at 6.3%", published 30 July 2026 for reference month June 2026; and the dataset behind it, une_rt_m, read on 27 August 2026 and stamped as updated on 25 August 2026. Nothing here comes from our own platform.
- One number is derived: the roughly 61,000 net rise outside the euro area, obtained by subtracting the euro-area change from the EU change (53,000 minus minus-8,000), and equivalently by subtracting the two published levels at each end. Eurostat publishes no non-euro-area subtotal. Both published figures are rounded to the nearest thousand, so the derived one carries their rounding.
- The composition question was checked, not assumed. Bulgaria adopted the euro on 1 January 2026, which would explain the divergence if the two ends of the comparison used different memberships. They do not: the euro-area series is published on the twenty-one-member composition for June 2025 as well as June 2026 (2,359,000 under-25s unemployed, identical in the release and in the current dataset), and queries for the twenty- and nineteen-member aggregates return no data.
- Romania has no monthly June figure. The release names it among five countries - with Belgium, Croatia, Cyprus and Slovenia - that report quarterly. Its last published under-25 rate is 28.8% for March 2026. The table therefore prints an empty cell rather than a number we estimated, and Sweden's 24.9% is described as the highest published rate for June 2026, never as the highest in the Union.
- Sweden's June rate was revised after publication: 24.9% in the release of 30 July, 25.8% in the dataset updated 25 August, with the all-ages rate moving from 8.7% to 8.9%. The table prints the release figure, because the release is what this piece reports. No other figure in the table changed between the two vintages, and neither did the minus 8 thousand or the plus 53 thousand.
- No share of the 61,000 is attributed to any country. The six outside states, as published for June 2025, sum to 584,000 unemployed under-25s against 575,000 implied by the two aggregates. We report the nine-thousand gap and do not explain it; rounded national counts and Romania's quarterly reporting are each sufficient to produce one, and we did not establish which applies.
- The under-25 rate divides by the youth labour force, not the youth population. No figure in this piece should be read as a share of all young people.
- Rates are seasonally adjusted, as published. Levels are as published, in thousands.
Primary sources
We publish how our own matching is tested for bias, with the score gaps it produces - the method and the numbers are here. A market that reports two opposite answers for the same year deserves tools that show their working.
