Where the Jobs Are Going
Demand, growth, decline and risk. The market half of the labor market.
The labor market in mid-2026 is frozen rather than collapsing. Job openings are roughly where they were in 2019, but there are more people competing for each one — 1.04 openings per unemployed person against 1.21 in 2019 — hires are down about 10 per cent, quits down 11 per cent, and layoffs are below their pre-pandemic average. Few people are being pushed out, and few are confident enough to move.
Almost everything written about the job market compares a number to a year without saying which year, and 2021 and 2022 were the strangest two years the data has on record. Pick them as your baseline and everything since looks like a collapse.
What the published series actually say
Openings around 7.36 million, slightly above the 2019 average. Hires about 5.35 million, roughly 10 per cent below it. Quits about 3.23 million, 11 per cent below. Layoffs about 1.77 million, below the 2019 norm.
That combination has a name worth using: low firing, low quitting, and fewer completed hires than advertising would suggest. Most openings are created by somebody leaving, so when people stop leaving, the vacancies stop appearing — which is why a frozen market sustains itself rather than correcting.
The number that settles the argument
Openings per unemployed person. It fell to about 1.04 from 1.21 in 2019, against a peak above 2.0 in early 2022. The openings count on its own flatters the picture because it counts vacancies without counting the people chasing them.
The longer arc
Over a decade the questions are different: which occupations are growing and shrinking, what automation is actually absorbing rather than what it could in principle, and where the work physically is. All three are answered with published data here, and all three are routinely answered badly elsewhere — by confusing percentage growth with job creation, exposure with displacement, and job count with job density.
Nobody can tell you what happens next
Employment projections are a model built on stated assumptions: full employment at the end of the horizon, no recession, no war. They are useful for direction and comparison, and they are not a promise about 2034.
The same goes for automation. Respected estimates of how much work is exposed range from 9 per cent to 47 per cent, and the gap is almost entirely about how the question was asked. Where the honest answer is a range or an open disagreement, you will find that here rather than a confident number.
In this guide
Automation and AI Exposure
10 articles →Growing and Shrinking Occupations
10 articles →Layoffs and Job Security
10 articles →Reading the Job Market
10 articles →Where the Jobs Are
10 articles →Common questions
Is the job market bad right now?
Frozen rather than collapsing. Openings are near 2019 levels, but openings per unemployed person fell to 1.04 from 1.21 and hires are down about 10 per cent.
Are layoffs unusually high?
No. At about 1.77 million a month they sit below the 2019 average and far below any recession, though that still annualizes to roughly 21 million separations a year.
Why does it feel worse than the data says?
Because expectations were set in 2021 and 2022, which were the anomaly. And a falling market feels different from a flat one even at the same level.
Which occupations are growing fastest?
Healthcare and data work. Nurse practitioners lead at 40.1 per cent over 2024-34, though percentage growth favors occupations starting from a small base.
Will AI take my job?
Published exposure estimates range from 9 to 47 per cent for the same question, and all of them measure what a technology could do rather than what employers actually do.
Does this guide predict the future?
No. Projections are a model under stated assumptions, and where the honest answer is a range or a genuine disagreement between sources, that is what is published.
What Jobs Pay
You know where the work is going. The first guide is how to find out what it pays.
Read it →