Britain lost 145,000 payrolled jobs this year. Almost exactly that many were people aged 25 to 34.

The tax records came out at seven this morning. Payrolled employment in Britain fell by 144,849 over the year to August. That is the number the bulletins will carry.

Underneath it is an age table, and the age table says something the headline cannot.

Payrolled employees aged 25 to 34 fell by 144,495. The total fall was 144,849. One age band accounts for 99.8 per cent of it.

That is not because everyone else held steady. Three age groups fell by 277,007 between them. Three rose by 132,159. Four hundred and nine thousand people moved across those bands in opposite directions, and what reaches the front page is the 145,000 left over when you add them up.

The group that grew fastest, by proportion, was people aged 65 and over, up 4.9 per cent.

Now the part I want near the top rather than buried at the bottom, because we leaned on this last week.

On Friday this paper told you the tax records showed under-25s leaving payrolled work in six figures. We cited 115,000 between March 2024 and March 2025 and we named that period. It was accurate. Today's release, same source, year to August 2026, shows under-25s down 15,366 in total, and inside that, people aged 18 to 24 grew by 26,270. The entire under-25 fall is now school-age workers, the 0 to 17 band, which is Saturday jobs rather than careers.

Nothing we printed was wrong. The ground has moved, the corroboration we relied on is weaker than it was four days ago, and you should hear that from us first.

Which brings me to what today's number does not tell you.

A fall in payrolled 25 to 34 year olds can mean two completely different things. It can mean fewer jobs for people that age. Or it can mean fewer people that age. Net migration to Britain was 204,000 in the year to June 2025, well down on preceding years, and migration lands hardest in exactly that band. A smaller population produces the same payroll fall with nobody's chances of work having changed at all.

We cannot separate those. Neither can anyone else, today, because the measure that would do it is the employment rate for that age group, and that needs a population count by single year of age which the ONS has not yet published for mid-2025.

I could write you a piece this morning saying AI is hollowing out the early career. The number would sit there and look like evidence. It is not evidence, it is a shape, and our own Contrarian named this exact objection four days ago when he listed a smaller cohort among the explanations that fit our evidence just as well. It would be poor form to apply his test to his own side of the paper and not to ours.

James has the arithmetic and the flows, which are the most interesting thing in the release. Ada has what a market like this feels like from inside it, which is not redundancy and is not recovery.

One thing worth holding on to while the rest is unresolved. The number of people leaving payrolled work rose 4.5 per cent over the year. The number arriving rose 0.7 per cent. Whatever is happening to the people in the middle of that, it is not a wave of sackings. It is a door that opens more slowly than it used to.

— M.

This note is mine: the view, and the call to run it. It begins as a draft, drawn from work the AI and I have researched and argued out together, the same way every desk in this paper is made, and I answer for every line because I read every line. Those desks run on models built by Anthropic, one of the labs sitting on the very scoreboard we report, so we tell you plainly: we cover this from inside it.