Signal vs. Noise
Low-Hire, Low-Fire: Safe If You're In It, Closed If You're Not
Two true statements about the same labor market, pointing at opposite strategies.
By Scot Free
This week's headline, from Fortune: AI is killing more than 10,000 jobs a month, and 9% of workers are at risk.
Also this week, from the Department of Labor: initial unemployment claims around 199,000, among the lowest readings in years.
Both are real. Neither is the thing you need to know.
What's measured
Start with the numbers that come from counting rather than modeling.
- Hiring is weak. Employers added 29,000 jobs in September, against a three-month average near 51,000. July was revised down to a loss of 10,000.
- Layoffs are not happening. Initial claims came in around 199,000 for the week ending October 10, against 222,000 in the comparable week a year earlier. Continuing claims sit near 1.73 million — low by any historical standard.
- Nobody is moving. The quits rate is holding at 1.9%. People are staying put, which is what workers do when they don't believe there's somewhere to go.
- And here is the number that should be the headline. Challenger, Gray & Christmas reports employers announced plans to hire 90,787 workers in September — the lowest September total since 2011.
Put those together and you get what Fed Governor Lisa Cook called a low-hire, low-fire equilibrium — the hiring rate is low, the unemployment rate stays steady because layoffs are also low, and the arrangement hits some groups, especially new entrants, particularly hard. A Vanguard economist put it more bluntly: it's bordering on a no-fire, no-hire market.
Now the forecast, checked
Back to the AI headline, because it's worth running the usual pass.
Pantheon Macroeconomics estimates AI is causing U.S. job losses exceeding 10,000 roles per month, and that under their baseline forecast of a 15% productivity increase from AI, around 9% of workers could be temporarily displaced during the transition.
Goldman Sachs says AI could displace one in 11 workers — but adds that the layoffs are likely waiting for a recession.
Four things about that pair:
- One in 11 is 9.1%. Pantheon says 9%. Two independent-looking forecasts landing on effectively the same figure is either genuine convergence or a shared input, and nobody reporting it asked which.
- Both are conditional. Pantheon's rests on a 15% productivity assumption. Goldman's rests on a recession that hasn't arrived. Strip the condition and you don't have a forecast, you have a sentence.
- "Temporarily" is doing quiet work. It's in Pantheon's wording and it vanishes in every headline.
- And the analysts themselves are hedging. One told clients that concerns of imminent AI-related job losses have cooled, while recent data shows some signs of AI-related hiring headwinds. Headwinds on hiring is a materially smaller claim than displacement, and it matches the measured data better.
None of this means AI isn't affecting employment. It means the mechanism is showing up as jobs not created rather than people walked out — which is consistent with every measured number above, and which is a quieter and harder problem to see.
Where it's landing
The sector split in the same analysis is the most actionable thing in the entire story:
Construction jobs have gone up — because tech companies are building data centers as fast as they can — while jobs in accounting, advertising, office admin, and legal services are slowly disappearing.
That is two of this site's arguments, arriving from a source that has never read either one.
The construction half is the Build-Out Board: the AI boom has a parts list, and somebody has to pour, wire, pipe and cool the buildings. September's 11,000 construction jobs and 17,000 health care jobs carried a month that added 29,000 total.
The legal services half is the paralegal finding: BLS projects that occupation flat through 2035 and names AI as the reason. An independent macro shop watching sector employment sees the same thing from the other end.
Two markets, two strategies
Here's why the distinction matters more than the headline.
If you have a job, this is among the safest markets in a decade. Claims are low, layoffs are low, and employers are hoarding the people they have because they remember how hard hiring was. Your downside risk is unusually small.
If you don't have one — or you want to move — it's the worst market since 2011 by the one measure that counts, which is whether anyone is planning to hire. And the people it hits hardest are exactly the ones with no incumbency to protect them: new graduates, first-time job seekers, and anyone trying to re-enter.
Those two situations call for opposite behavior, and most advice collapses them into one.
- Employed: this is the year to build the case inside the building, not outside it. The self-review, the calibration room, the rung above you. Internal moves don't require anyone to open a req, and the quits rate at 1.9% means your competition for that rung is thinner than usual.
- Looking: postings are the worst channel they've been in years, because employers continue signaling demand through job postings without following through with hiring. That's an economist's description of a req that exists on a board and not in a budget. Applications into that are 200 unmeasured trials.
- Either way: the premium is narrow and specific. Average new-hire salary growth across seven professional fields is projected at 1.9% for 2027 — but 57% of employers say they're paying above their planned range for particular skills. Flat market, pointed exceptions.
When does it break?
The honest answer is that nothing measured says it improves before the middle of next year.
- Rates moved the wrong way. After three cuts in late 2025, the Fed reversed and raised in September, with another hike expected before year-end and cuts not resuming until late 2027 on one major forecast. Goldman has pushed its cut forecast to June and December 2027.
- The forward measurement is flat. Indeed projects job postings falling 1.4% through June 2027, with unemployment near 4.4%.
- Stated intent is rosier, as always. 31% of employers expect to grow headcount in 2027 against 9% expecting cuts. Treat that the way you'd treat any survey of intentions — directionally interesting, historically optimistic.
- And part of this isn't cyclical. The working-age population is projected roughly unchanged across 2026 and 2027. Some of "low-hire" is that there are fewer people to hire, and a rate cut doesn't fix demographics.
So: don't wait for it to turn. Plan on this being the shape of the market through at least mid-2027, and make the moves that work inside it.
The Scot Free Take
The AI displacement number got the headline and the lowest September hiring plans since 2011 didn't, and the second one is the number that actually describes somebody's week.
What strikes me is how badly "the job market is fine" and "the job market is terrible" both perform as descriptions. Low layoffs and no hiring are the same market, and whether it's good or bad depends entirely on which side of the door you're standing on. Nobody writes it that way because it doesn't fit a headline.
The practical read: if you're in, use the year to move up inside. If you're out, stop treating postings as the market — a req on a board that isn't in a budget is not an opening, and the measured data says a lot of them aren't.
Get the MoneyZoo LAB Report and instant access to The Side Door Playbook. One seat, one receipt, one move, every Monday. Subscribe here.
Sources
- Fortune, Oct. 23, 2026 — Pantheon Macroeconomics (Samuel Tombs, Oliver Allen) on AI-related job losses exceeding 10,000 per month and roughly 9% of workers temporarily displaced under a 15% productivity baseline; the construction-versus-office-sectors split; the analyst note that concerns of imminent AI job losses have cooled while data shows AI-related hiring headwinds. Goldman Sachs "one in 11" displacement estimate with layoffs likely waiting for a recession.
- BLS Employment Situation, released Oct. 2, 2026 — 29,000 jobs added in September; unemployment 4.2%; health care +17,000, construction +11,000, manufacturing +9,000, government −17,000; July revised to −10,000; three-month average near 51,000. Next report scheduled Nov. 6.
- U.S. Department of Labor weekly unemployment insurance claims, weeks ending Oct. 3 and Oct. 10, 2026 — initial claims near 197,000–200,000, against 222,000 in the comparable 2025 week; continuing claims approximately 1.71–1.73 million. JOLTS quits rate 1.9%.
- Challenger, Gray & Christmas — 90,787 announced hiring plans in September 2026, the lowest September total since 2011. Note that Challenger tracks announcements rather than completed actions.
- Federal Reserve Governor Lisa Cook on the low-hire, low-fire equilibrium; Adam Schickling, Vanguard, on a "no fire, no hire" market; Cory Stahle, Indeed Hiring Lab, on employers signaling demand through postings without following through on hiring.
- Deloitte US economic forecast (Oct. 2026) — the Fed's September rate increase, an expected further hike before year-end, cuts resuming late 2027, and flat working-age population across 2026–27. Goldman Sachs Research — rate cuts pushed to June and December 2027.
- Indeed — job postings projected to fall 1.4% through June 2027, unemployment near 4.4%. The Conference Board — 31% of employers expecting headcount growth in 2027 against 9% expecting cuts. Robert Half 2027 Salary Guide — 1.9% average projected new-hire salary growth across seven fields; 57% paying above planned range for specialized skills.
A note on these sources: the measured series here — payrolls, claims, quits — come from federal statistical programs. The displacement figures are private forecasts with stated conditions attached, the Challenger figure counts announcements rather than completed hires, and the employer intent and salary figures come from survey work by firms with a commercial interest in hiring activity. All are labeled above as what they are.
