Jobless claims rose 9,000 last week on the adjusted measure, and the raw count fell to below where it stood a year ago
The headline says claims rose. The unadjusted table says they fell.
Both are in Thursday's release, and the second one explains the first. Seasonally adjusted initial claims came in at 197,000 for the week ending 25 July, up 9,000 on a previous week revised to 188,000. Underneath, the actual count of first-time filings under state programmes was 175,573, down 17,803 from the week before, a fall of 9.2 percent.
The seasonal factors had expected a fall of 13.3 percent. Claims did drop. They simply did not drop as far as late July usually makes them drop, and the adjustment turns that shortfall into an increase.
The comparison that needs no adjusting
There were 193,790 initial claims in the same week of 2025. This year the raw figure is 175,573, which is 18,217 lower.
The continuing side reads the same way. The insured unemployment rate held at 1.2 percent for the week ending 18 July, against 1.3 percent a year earlier. Seasonally adjusted insured unemployment fell 7,000 to 1,782,000, where the year-earlier figure was 1,936,000. Counting every programme, 1,872,139 weeks of benefits were claimed in the week ending 11 July, against 2,037,427 in the comparable week last year. No state was triggered on to the Extended Benefits programme.
The four-week moving average of initial claims, which exists precisely to absorb weeks like this one, fell 5,000 to 202,750.
One state, and then a long way down
The state table lags the headline by a week, so it covers the week ending 18 July. In that week the release lists five states with increases in initial claims: Louisiana at 346, Delaware at 162, Vermont at 75, West Virginia at 12, and Wyoming at one.
Wyoming's increase was a single claim.
Against that, New York fell 17,156, Michigan 4,974, California 4,242, Texas 2,354 and Pennsylvania 2,266. The largest increase in the country was smaller than the smallest listed decrease by a factor of six.
Claims count layoffs. They have never counted hiring, and a report this quiet on the layoff side says nothing at all about how long it takes someone already out of work to find the next job.