Jobless claims fell to 197,000 — almost nobody is being fired, and that keeps a December rise in play.

Jobless claims count how many people filed for unemployment benefit last week. It is the fastest available measure of how many people are losing their jobs.

The call

The number. 197,000 new claims, against 200,000 expected and 199,000 the week before. Near a 57-year low.

What that does. Firms are not letting people go. But they are not hiring either — September payrolls came in at just 29,000. That is a frozen jobs market, not a collapsing one, and a frozen market does not force a central bank to stop.

The rate. 3.75–4.00%, raised on 16 September. October looks unlikely; around 80% is priced for a rise by December.

The currency. Higher rates make a currency worth more to hold, and this removes the strongest argument against December. Dollar supported.

The print

Initial claims 197,000, against 200,000 expected. The previous week was revised up to 199,000.

Four-week average 198,000 — the lowest since October 2022, and a fourth straight week below 200,000.

Continuing claims 1,716,000, up 17,000. The insured unemployment rate held at 1.1%.

Why it matters

The weak September payroll figure raised a real question: is the American labour market breaking? Claims answer it, and the answer is no.

Firing is the fast signal, and there isn’t any. Claims have sat below 200,000 for a month and the four-week average is at a four-year low. A labour market in trouble shows up here first, and it hasn’t.

What has changed is the other side. Hiring has stalled, continuing claims are drifting higher, and the median spell of unemployment is now 11.5 weeks — close to a four-and-a-half-year high. So people are not being pushed out, but anyone who does lose a job is taking far longer to find another. Heather Long of Navy Federal calls it a “low-hire, low-fire” market.

Companies are sitting still. Tariffs, the war with Iran and record diesel prices have made firms reluctant to commit to new staff, while a smaller workforce from retirements and immigration restrictions has reduced the pressure to let anyone go.

The case

For the Fed, the distinction is everything. A central bank will abandon a rate rise if unemployment is climbing because people are being sacked. It will not abandon one because hiring has gone quiet while inflation is still too high.

This print protects the December case. It does not create a new one — the rate was set three weeks ago and the next decision is 28 October, which almost nobody expects to produce a move.

Ideas

Direction only, no entry here. One weekly figure, three thousand below forecast, is not a reason to buy the dollar. It is a reason not to sell it. The dollar’s next real test is the 28 October decision and the inflation data before it.

Sources

US Department of Labor, weekly claims release, 8 October 2026 · Vantage — claims near 57-year low, duration, December pricing · Babypips — low-hire low-fire, labour supply, Fed context