Skip to content
BabZıtunaThe Market Desk

The record · 17 August 2026

One unemployed American per job opening. That is not full employment.

In June the United States counted 1.0 unemployed person per job opening - and it has counted 1.0 or 1.1 for eight consecutive readings. Across the Atlantic, the ratio this desk reported yesterday stands at 2.5. Both numbers divide a stock of jobless people by a stock of open posts; neither counts applications, employed applicants, or anyone outside the labour force. And a one-to-one ratio did not clear the market: the same June release shows 7.4 million openings, 5.3 million hires, 3.2 million quits and 1.8 million layoffs, every one of them described by the agency itself as little changed. Every figure below is from a published source, listed and linked at the foot of this piece.

1.0unemployed Americans per job opening, June 2026 - a ratio of stocks, not a queue of applicants
7.4Mjob openings on the last business day of June 2026, per the BLS - little changed on the month
2.5the British ratio this desk reported yesterday - a different instrument, quoted for contrast, not comparison

What a 1.0 actually counts

The Bureau of Labor Statistics publishes a ratio that answers one narrow question with unusual precision: how many unemployed people are there for each open job? For June 2026 the answer is 1.0 - and it has been 1.0 or 1.1 at every reading since November 2025. The numerator counts people classified as unemployed: out of work, available, actively looking. The denominator is the JOLTS count of job openings - and that word has a definition worth quoting, because everything else in this piece leans on it. A position counts as open only if it exists on the last business day of the month, could start within 30 days, and the employer is actively recruiting for it.

Note what is missing, because it is the same list as yesterday. Applications are in neither term. One unemployed person sends dozens; the ratio counts them once. Employed applicants are in neither term - in most markets a large share of every applicant pool already has a job and is, by definition, not unemployed. And everyone outside the labour force is outside the arithmetic entirely.

So the ratio is exactly what it says and nothing more: a stock of jobless people set against a stock of advertised work. In America right now those stocks are the same size. What that equality means - and does not mean - is the rest of this piece.

Five things people mean when they say "how is the American job market". They come from one release, they are not versions of one number, and they will not move together.
MeasureJune 2026What it actually answers
Unemployed persons per job opening1.0How many jobless people share each open post
Job openings7.4MPosts open on the last business day of the month
Hires5.3MPeople actually added to payrolls in the month
Quits3.2MVoluntary departures - workers choosing to leave
Layoffs and discharges1.8MInvoluntary separations - employers letting go

One-to-one is not a handshake

If every unemployed American could be walked to a matching opening, a ratio of 1.0 would end unemployment by arithmetic. The same release shows why that sentence is fantasy. In the month the ratio read one-to-one, employers hired 5.3 million people - and 7.4 million openings still sat unfilled at month's end. Positions and people coexist without meeting: the opening is in one state, the person in another; the post wants a licence the person does not hold; the shift does not survive contact with a school run.

The June release makes a second point that is easy to miss because it is phrased as an absence. The BLS describes every headline series as unchanged or little changed - openings, hires, total separations, quits, layoffs. Eight readings of a flat ratio say the same thing from the other side. This is not a market in turmoil; it is a market that has gone still. Hiring has neither collapsed nor recovered; employers are neither purging nor stockpiling. The one-to-one ratio is the arithmetic of that stillness, not a door held open.

For a job seeker, stillness cuts both ways. 3.2 million Americans quit voluntarily in June - people do move, and every move opens a seat. But a still market rewards precision over volume: when hiring is flat, the marginal application lands on a desk that is in no hurry. The advantage shifts to applications that arrive matched, complete, and early - which is not a slogan; it is what a flat hires line arithmetically implies about the queue behind each post.

The Atlantic is a definition wide

Yesterday this desk reported the British ratio at 2.5 unemployed people per vacancy. Today's American figure is 1.0. It is tempting to print those as one series with an ocean in the middle, and it would be wrong in a way worth being precise about: a JOLTS "job opening" and an ONS "vacancy" are built by different agencies, from different surveys, under different rules. JOLTS requires the last-business-day snapshot, the 30-day start, the active recruitment. The ONS counts from a survey of businesses with its own frame and exclusions. The words are cousins, not twins.

Having said that clearly, the honest next sentence is: no definitional quibble closes a gap of two and a half to one. Britain's jobless stock is spread two and a half deep across its advertised posts; America's is spread one deep. The direction of that difference is real, survives every reasonable adjustment, and is the reason this piece and yesterday's are a pair.

What the pair teaches is about ratios, not geography. Yesterday's number looked alarming and turned out to describe a plateau reached by both terms shrinking together. Today's looks reassuring and turns out to describe a stillness in which millions of openings and millions of jobless people sit side by side without touching. Neither ratio, read alone, told you what its market feels like. That is not a flaw in the statistics; it is a property of quotients, and the reason a desk that quotes one should always show its terms.

The BLS series itself, eight readings, every value quoted from the agency's published data table - none derived. Bars from zero; the axis runs to 2.5, the British reading, so the gap between the two markets is the white space. A series this flat is a finding: the American ratio has not moved meaningfully in eight months.
MonthUnemployed per opening
November 20251.1
December 20251.1
January 20261.0
February 20261.1
March 20261.1
April 20261.0
May 20261.0
June 20261.0

The revision, again - and what to do inside a still market

Four days before this piece, the BLS revised May: openings down 57,000, hires up 82,000, total separations up 159,000. Readers of this desk's 13 August edition on first-draft statistics will recognise the pattern - the first print moved, in three directions at once, by amounts that would each have made a headline. None of the revisions changed the shape of the story, which is precisely the point of reading levels as drafts and trends as the record.

The practical reading for a candidate in the American market is neither "jobs are everywhere" nor "hiring is frozen". It is narrower: openings exist at scale, and they are filled slowly. 5.3 million hires a month is an enormous absolute number that is nonetheless flat - so the posts that do fill, fill against a standing queue. In that market the scarce asset is not the opening; it is the application that fits - the one aimed at the right post, complete on arrival, sent while the post is fresh. That, and not application volume, is what a flat one-to-one market pays for.

And for anyone hiring: your competitors are neither cutting nor hoarding. 1.8 million involuntary separations a month against 3.2 million voluntary quits means churn is running on workers' decisions more than employers'. The candidate you want is statistically likelier to be leaving somewhere than to be unemployed - which is an argument for speed between first contact and offer, because employed candidates do not wait out a still market's hiring pace.

We publish how our own matching is tested, including where it is weakest, at our bias audit. If a number here is wrong, that page is where the correction should start.

Earlier editions