Job Market Disconnect
Why So Many Job Openings But So Little Hiring?
Job openings say one thing about the labour market. Hiring speed and how long people stay unemployed say another. This measures how far apart they are.
June 2026 reading · 294 comparable months · data retrieved 2026-08-22
Current disconnect
+46
percentile points · 98th percentile of 294 months
- Openings
- 23rd percentile
- Hiring velocity
- 69th percentile
- Search duration
- 69th percentile
(69 + 69) ÷ 2 − 23 = 46
Every percentile above is a difficulty ranking against 294 comparable months: higher means harder. The arithmetic is shown because a measure nobody can check by hand is a measure nobody should repeat.
What is the Job Market Disconnect?
The Job Market Disconnect measures how far apart the job-openings signal and the hiring-and-duration signal are in any month. In June 2026 the gap is 46 percentile points — the 98th percentile of 294 comparable months since 2000. Job openings rank in the easier 23rd percentile of difficulty, while hiring velocity ranks at the 69th and how long people stay unemployed at the 69th. Vacancies are plentiful by historical standards; hiring and search length are not keeping pace with them.
How unusual is it?
Across 294 comparable months the gap has run from -53 to 51, with a median of 0.5 — for most of this century the two signals broadly agreed. The current reading is in the top 2 per cent of that distribution.
Why it is not just the Friction Index again
It correlates with the Job Search Friction Index at r = 0.037, which is close enough to zero to be independent of it. The index says how hard conditions are. This says whether the indicators agree about it.
History
Every month since 2000.
Positive means hiring and search look worse than the supply of openings implies. Negative means the reverse — hiring better than the vacancy count suggested.
Reading this: above the line, hiring and search duration look worse than the supply of openings implies. Below it, better. The break is March 2020 to February 2021, excluded as structurally unreliable.
The series runs from -53 in October 2001 to +51 in May 2026, and currently reads +46.
When it crossed
- First month above +20
- July 2015 · 58 months total
- First month above +30
- August 2021 · 33 months total
- First month above +40
- June 2024 · 12 months total
- First month above +45
- November 2024 · 6 months total
Longest sustained periods
- Above +20
- July 2023 → June 2026 · 35 months
- Below −20
- July 2003 → June 2008 · 60 months
- Widest month on record
- May 2026 · +51
- Most negative month
- April 2002 · -53
The long drift
This did not start in 2024.
The month-to-month reading is noisy. The decade averages are not: the relationship has been moving in one direction for twenty-five years.
The gap between the two signals has inverted
Average gap in percentile points, by period · n = 294
no months above +40
no months above +40
no months above +40
no months above +40
no months above +40
12 months above +40
Source data: BLS JOLTS and Current Population Survey
Analysis: Jobsearch.ing
Last updated: 2026-08-22
Bars run from a shifted origin because the earlier periods are negative; the figure on each bar is the true signed gap.
By decade
- 2000s
- mean -30.4
- 2010s
- mean +6.7
- 2020s
- mean +28.8
109 months · -53 to -6.5
120 months · -12 to 26.5
65 months · 7.5 to 51
In the 2000s hiring was consistently better than the vacancy count implied. In the 2020s it is consistently worse. The sign flipped during the 2010s, long before the current readings.
Testing the obvious story
Have job openings stopped predicting hiring?
The natural conclusion is that the openings number has broken as an indicator. We tested that, and it is not what happened.
If openings were losing their informational value, the two signals would be drifting apart in their movements — the correlation between them inside each period would fall. It does not. Across every five-year block, including the most recent, they move together between r = 0.87 and r = 0.985.
What changed is the level. The opportunity signal fell 49.5 percentile points between the early 2000s and today — openings became abundant by historical standards — while hiring and duration moved up just 15.4.
So openings still tell you which way hiring is going. What they no longer tell you is how good the hiring is: the same vacancy abundance that once accompanied a fast market now accompanies a middling one.
| Period | r | Openings | Hiring |
|---|---|---|---|
| 2001–2005 | 0.87 | 64.1 | 28.6 |
| 2006–2010 | 0.98 | 71 | 51 |
| 2011–2015 | 0.94 | 72.2 | 75 |
| 2016–2020 | 0.89 | 25.2 | 39.2 |
| 2021–now | 0.90 | 14.6 | 44 |
Openings and Hiring columns are mean difficulty percentiles. Lower means easier. The correlation column is within-period, so it measures co-movement rather than level.
Explanations
What could explain the disconnect?
Separated by what these data can actually carry. The first list is arithmetic. The second is a list of candidates we cannot choose between.
Directly measured
Openings are abundant relative to this century's history.
The opportunity signal averaged 14.6 on a difficulty percentile since 2021, against 64.1 in the early 2000s. Lower means easier.
Hiring and search duration have not followed.
Over the same period they averaged 44, against 28.6 in the early 2000s — a much smaller move, and in the opposite direction to the openings signal.
The two signals still move together.
Within every five-year block the correlation between them sits between 0.87 and 0.985. Openings still track hiring; they simply sit at a different level relative to it.
The gap is not a restatement of the headline index.
It correlates with the Job Search Friction Index at r = 0.037.
Plausible hypotheses, not established here
Postings staying open longer
A vacancy that remains advertised without being filled is counted in every month it is open, so a slower process inflates the stock of openings without any change in hiring.
Changes in how vacancies are posted
Cheaper posting, duplicate listings across platforms, and speculative advertising would all raise measured openings relative to actual hiring.
Slower or more selective hiring processes
More interview rounds, more approvals, or a higher bar for filling a role would lengthen searches while leaving the vacancy count untouched.
Mismatch between advertised roles and available workers
Skills, seniority, location, or pay expectations that do not line up would leave openings unfilled and searches long at the same time.
Composition of who is unemployed
A shift in which industries and which workers are searching would change duration without changing the supply of vacancies.
Each of these would produce the pattern we measure. Nothing in these series distinguishes between them, and we have not seen a dataset that does.
Not established by this data
- Which of the hypotheses above is responsible, or in what proportion.
- Whether employers are advertising roles they do not intend to fill.
- Whether the change is permanent or a phase of the current cycle.
- Whether any particular industry, occupation or region is driving it.
- Whether job seekers face worse odds per application than they used to.
Sensitivity
Does the finding depend on how we defined it?
Four ways of asking the same question. The divergence is real under all of them. One striking claim is not.
| Definition | Current | Percentile | Months > +40 | Before 2024 |
|---|---|---|---|---|
| Mean of hiring and durationpublishedThe published measure. | 46 | 98th | 12 | 0 |
| Hiring onlyDrops search duration entirely. | 46 | 92nd | 31 | 3 |
| Duration onlyDrops hiring velocity entirely. | 46 | 95th | 15 | 12 |
| Worse of the twoTakes whichever signal looks harder, rather than averaging. | 46 | 88th | 43 | 15 |
Read this before quoting the striking number
Under the published measure, no month in the 294 comparable months before June 2024 exceeded +40. That is true, and it is a property of this construction, not of the labour market: under a duration-only definition, 12 earlier months cross the same line. The divergence itself is present in every version. The “never before” framing is not, so we would rather you quoted the percentile.
Method
How it is calculated.
Each of the three dimensions of the Job Search Friction Index is converted to a difficulty percentile against 294 comparable months. The disconnect is the average of the hiring and search-duration percentiles minus the opportunity-pressure percentile.
The percentiles used are the rounded values shown on this page, so the published figure is exactly what a reader reproduces by doing the subtraction themselves. March 2020 to February 2021 is excluded, for the same reason the index excludes it: a mass layoff and rehire inverts the components.
No new data is collected for this measure. It is a different view of the same series, which is why it can be checked against the index rather than taken on trust.
Cite this
Jobsearch.ing, “Job Market Disconnect”, June 2026. 46 percentile points, 98th percentile of 294 months. Source data: U.S. Bureau of Labor Statistics JOLTS and Current Population Survey, retrieved 2026-08-22. https://jobsearch.ing/job-market/job-market-disconnect//