Jobless Claims 206,000 (vs 205K Est) — The Four-Week Average Turned Back Down

Fundamentals · 2026-09-10

Initial claims 206,000 for the week ending September 5, down 1,000 against a 205,000 consensus, with the prior week revised up 1,000 to 207,000; the 4-week average fell 1,500 to 206,000, its first decline in five weeks, ending the climb that began at the August 1 low; insured unemployment 1,774,000, better than the 1,780,000 expected, its 4-week average down a third straight week to 1,779,000; unadjusted claims rose 5,164 where seasonal factors expected 5,789 — a 625 gap, the quietest mechanics in a month; exactly one state posted an increase above 1,000 — New York, +4,338 — and named transportation and warehousing, health care and social assistance, and educational services; work-sharing 45.2 percent below a year ago; the four-week average is 13.9 percent under last year.

What Is This?

• What it is: The Labor Department's weekly count of unemployment insurance activity, released Thursdays at 8:30am ET. Initial claims count people filing after a separation from an employer — the cleanest high-frequency read on layoffs anywhere in the calendar. Insured unemployment, or continued claims, counts weeks of benefits actually claimed and lags initial claims by a week. These are administrative counts from every state program office rather than a survey, so there is no sampling error, but weekly seasonal adjustment is difficult and the series is volatile.
• Why it matters: It shares its morning with producer prices, sits one day ahead of the August consumer price report, and lands five days before an FOMC at which three officials dissented in July in favor of a hike. It is also the release that carries a direct correction to last week's reading of it: seven days ago the four-week average had risen four weeks running, and that was the first sign in this cycle that the layoff channel had stopped improving. This week it did not.
• How to read it: The weekly print is noise; the four-week average is the trend. Read the unadjusted number against what the seasonal factors expected, because that gap — not the raw move — is what drives the headline. And hold the distinction firmly: claims measure firing, not hiring. A low number says people are not being let go. It says nothing about whether anyone is being taken on.

Want the full explainer? Economic News Events, Explained breaks down this release and every other one we cover, in plain English.

Summary

Initial claims came in at 206,000 for the week ending September 5, down 1,000 from a prior week revised up 1,000 to 207,000. Consensus was 205,000, so the print landed a thousand above it — inside the rounding of a series that moves in tens of thousands. Insured unemployment was the cleaner beat, at 1,774,000 against an expected 1,780,000, with the prior week revised down 4,000 to 1,775,000.

The trend is what changed. The four-week moving average fell 1,500 to 206,000, its first decline in five weeks. It bottomed at 199,000 in the week ending August 1, then rose four consecutive weeks to 207,500. That run is over.

This is a direct update to what this feed wrote a week ago. Last Thursday the rising four-week average was the finding, and it was set against Kevin Warsh, who had cited "unemployment claims, on a four-week average" at Jackson Hole on August 28 as an "empirically robust real-time indicator" sitting near its lowest level in decades. The point made here was that the specific measure he named had risen every week since he named it. It has now fallen. His characterization holds and the direction argument is a week weaker than it looked.

It is not erased. The four-week average at 206,000 is still 7,000, or 3.5 percent, above the August 1 low. One down week inside a five-week drift is a pause, not a reversal, and anyone treating it as either is reading a series that has spent the last five weeks inside an 8,000-claim band: 212, 207, 204, 207, 206.

The mechanics underneath were the quietest in a month, and that matters because the mechanics have been doing all the work. Unadjusted claims rose 5,164, or 3.0 percent, to 176,567, where the seasonal factors expected an increase of 5,789, or 3.4 percent. The raw data beat the seasonal norm by 625 — which is why the headline fell despite raw claims rising. Compare that to the two prior weeks: raw claims beat the norm by 3,024, then missed it by 1,256. Both of those were the story of their release. This week there is nothing to explain away.

The continued-claims side went marginally the other way. Unadjusted insured unemployment fell 58,476, or 3.4 percent, to 1,675,966, where the factors expected a fall of 56,809, or 3.3 percent — so the raw data underperformed the seasonal norm by 1,667, which is why the adjusted level only fell 1,000. The seasonally adjusted insured unemployment rate held at 1.2 percent and the unadjusted rate held at 1.1 percent, both unchanged, against 1.3 and 1.2 percent respectively a year ago. The denominator is covered employment of 153,732,307.

The two four-week averages have now converged. Initial claims rose four weeks and fell this one; continued claims have fallen three consecutive weeks, from 1,786,750 to 1,779,000. A week ago they were moving in opposite directions, which was worth flagging. They are now both pointing down, which is the more benign configuration and the less interesting one.

The state detail is where this release earns its reading. For the week ending August 29, exactly one state posted an increase above the 1,000 reporting threshold: New York, +4,338. No state posted a decrease above 1,000. That is the first state comment in three weeks — the two prior releases had none in either direction — and New York's explanation is the sentence to take out of this report: "Layoffs in transportation and warehousing, health care and social assistance, and educational services industries."

Read that against the last fortnight. Transportation and warehousing is the sector this morning's producer price report put at 13.0 percent over twelve months, the hottest major line in it, and the sector the preliminary benchmark revision published on August 28 said had employment understated by 135,100 — the largest upward correction in the table. Health care and social assistance plus educational services is private education and health services, the sector that accounted for more than the entire private payroll gain in ADP's August report, the sector carrying essentially every recent payroll print, and the one the same benchmark said was overstated by 96,000. The single named layoff cluster in the country this week hit both the fastest-inflating sector and the only sector holding the employment numbers up.

One state's weekly comment is not a trend, and this needs saying plainly: New York's spike had already unwound by the following week. In the advance data for the week ending September 5, New York fell 3,571 — from 18,237 to 14,666 — the largest state decline in the country and almost exactly a reversal. Whatever happened in New York was a one-week event in the data. What is worth carrying forward is not the level but the attribution.

The advance week's increases came from elsewhere: California +2,221 to 37,642 and Michigan +2,149 to 5,876, the only two states above 1,000 on the upside, followed by Washington +935, Oregon +737, Massachusetts +579 and Nebraska +565. Michigan is the one to note. It swung 2,446 lower three weeks ago on manufacturing layoffs, sat at a trivial +229 last week, and has now risen 58 percent in a week off a low base. These are unadjusted state figures on an advance basis, revised in seven days and reported by state of liability rather than residence, so no single week carries much — but Michigan and California moving together is a goods-and-technology combination rather than a services one.

Beneath the headline, the composition of who is claiming continues to shift in a way the totals hide. Work-sharing — short-time compensation, where an employer cuts hours across a workforce instead of laying people off — fell 543 to 14,208, and sits 45.2 percent below the 25,912 of a year ago. That gap has widened every week this feed has tracked it: 37 percent, then 39.1 percent, now 45.2. Employers are not reaching for the tool that normally precedes layoffs, which is the single most reassuring number in the release.

Against that, State Additional Benefits — the programs for claimants who have exhausted regular and, where applicable, extended benefits — rose 216 to 3,147, and are up 31.7 percent against a year ago. Total continued claims across all programs fell 8.7 percent over the same period. So the one category that is growing is the one that counts people who have run out of ordinary benefits. The number is tiny — 3,147 claims, 0.18 percent of the 1,758,129 total — and a single state's administrative decision could move it, so it proves nothing on its own. It is exactly what a low-fire, low-hire labor market would produce: fewer people entering the system, and the ones who do entering for longer.

Newly discharged veterans remain the other exception. Initial claims from veterans were 409, down 4 on the week but up 29.0 percent against 317 a year ago, and their continued claims of 4,902 are up 5.0 percent against 4,667. Former federal civilian employees moved the opposite way — 388 initial claims, up 51 on the week but down 26.4 percent year over year, with continued claims down 30.3 percent. The federal workforce disruption is normalizing. The veteran series has not, and it is now several weeks running.

On the level, nothing here is a distress signal. Initial claims are 13.9 percent below last year on the four-week average, insured unemployment 7.9 percent below, continued weeks claimed across all programs 8.7 percent below at 1,758,129 against 1,925,338, and no state anywhere is triggered on for Extended Benefits. At 206,000 the print sits below the 2026 average of roughly 210,600 and comfortably inside the year's 189,000-to-230,000 range. The layoff channel is closed.

What it does not tell you is whether anyone is hiring, and every other labor series this fortnight has said they are not: JOLTS hires down 278,000 with implied net employment of 18,000 lost, ADP private payrolls at 38,000 with one sector accounting for more than the whole gain, ADP's weekly pace 61 percent below its June level, and a benchmark revision that took 178,000 private jobs off the count through March. Claims remains the one series that has stayed clean. This week it stayed clean and stopped deteriorating at the same time, which is the best it has looked since the start of August — and it still measures only half the labor market.

The Internals

Initial claims, seasonally adjusted and not:

Measure · Sep 5 · Aug 29 · Change · Aug 22 · Year ago

Initial claims, seasonally adjusted · 206,000 · 207,000 · Down 1,000 · 204,000 · 259,000

Initial claims, unadjusted · 176,567 · 171,403 · Up 5,164 · 170,596 · 204,862

4-week moving average, adjusted · 206,000 · 207,500 · Down 1,500 · 205,750 · 239,250

Insured unemployment, one week in arrears:

Measure · Aug 29 · Aug 22 · Change · Aug 15 · Year ago

Insured unemployment, adjusted · 1,774,000 · 1,775,000 · Down 1,000 · 1,771,000 · 1,927,000

Insured unemployment, unadjusted · 1,675,966 · 1,734,442 · Down 58,476 · 1,771,025 · 1,803,347

4-week moving average, adjusted · 1,779,000 · 1,780,750 · Down 1,750 · 1,786,750 · 1,940,750

Insured unemployment rate, adjusted · 1.2 percent · 1.2 percent · Unchanged · 1.2 percent · 1.3 percent

Insured unemployment rate, unadjusted · 1.1 percent · 1.1 percent · Unchanged · 1.2 percent · 1.2 percent

Against consensus:

Measure · Actual · Consensus · Result

Initial claims · 206,000 · 205,000 · 1,000 above, effectively in line

Insured unemployment · 1,774,000 · 1,780,000 · 6,000 below, a beat

Insured unemployment rate · 1.2 percent · 1.2 percent · In line

The seasonal mechanics, which are what actually produced the headline:

Series · Raw change · Seasonal factors expected · Gap · Effect on the adjusted number

Initial claims, week to Sep 5 · Up 5,164, or 3.0 percent · Up 5,789, or 3.4 percent · 625 better than expected · Adjusted level fell 1,000

Insured unemployment, week to Aug 29 · Down 58,476, or 3.4 percent · Down 56,809, or 3.3 percent · 1,667 worse than expected · Adjusted level fell only 1,000

The four-week average, week by week, off the August low:

Week ending · Initial claims · 4-week average · Direction of the average

August 1, 2026 · 200,000 · 199,000 · The cycle low

August 8, 2026 · 212,000 · 199,750 · Up 750

August 15, 2026 · 207,000 · 204,250 · Up 4,500

August 22, 2026 · 204,000 · 205,750 · Up 1,500

August 29, 2026 · 207,000 · 207,500 · Up 1,750

September 5, 2026 · 206,000 · 206,000 · Down 1,500

Year-over-year, every headline series remains well below 2025:

Measure · Latest · Year ago · Change

Initial claims, weekly · 206,000 · 259,000 · Down 20.5 percent

Initial claims, 4-week average · 206,000 · 239,250 · Down 13.9 percent

Insured unemployment · 1,774,000 · 1,927,000 · Down 7.9 percent

Insured unemployment, 4-week average · 1,779,000 · 1,940,750 · Down 8.3 percent

Continued weeks claimed, all programs · 1,758,129 · 1,925,338 · Down 8.7 percent

One caution on that first line. The comparable week of 2025 printed 259,000 after jumping 23,000 in a single week — it was itself an outlier, which flatters the 20.5 percent. The four-week comparison at 13.9 percent is the honest one.

State And Program Detail

For the week ending August 29, one state cleared the 1,000 threshold in either direction:

Direction · State · Change · State-supplied comment

Increase above 1,000 · New York · Up 4,338 · Layoffs in transportation and warehousing, health care and social assistance, and educational services industries

Decrease above 1,000 · None · Not applicable · No state qualified

The rest of that week's movement, none of it large enough to require an explanation:

Largest increases · Change · Largest decreases · Change

New York · Up 4,338 · New Jersey · Down 814

Hawaii · Up 475 · Ohio · Down 545

Arkansas · Up 287 · Pennsylvania · Down 510

Florida · Up 253 · Michigan · Down 472

Rhode Island · Up 231 · Illinois · Down 401

The advance week, ending September 5, unadjusted and by state of liability. Only two states rose more than 1,000, and the largest single move in the country was New York giving back its spike:

State · Advance, week to Sep 5 · Prior week · Change

California · 37,642 · 35,421 · Up 2,221

Michigan · 5,876 · 3,727 · Up 2,149

Washington · 5,277 · 4,342 · Up 935

Oregon · 4,515 · 3,778 · Up 737

Massachusetts · 4,527 · 3,948 · Up 579

Nebraska · 1,056 · 491 · Up 565

Connecticut · 3,216 · 2,773 · Up 443

New Jersey · 8,151 · 7,723 · Up 428

Arkansas · 1,079 · 1,451 · Down 372

New York · 14,666 · 18,237 · Down 3,571

Highest insured unemployment rates for the week ending August 22:

State · Insured unemployment rate · State · Insured unemployment rate

New Jersey · 2.6 percent · Oregon · 1.9 percent

Puerto Rico · 2.6 percent · Washington · 1.9 percent

Rhode Island · 2.2 percent · California · 1.8 percent

Massachusetts · 2.1 percent · Connecticut · 1.7 percent

Minnesota · 1.9 percent · Nevada · 1.7 percent

New York · 1.7 percent · National rate, unadjusted · 1.1 percent

Continued weeks claimed across every program, week ending August 22. This is the layer the headline hides:

Program · Aug 22 · Aug 15 · Change · Year ago · Year-over-year

Regular state programs · 1,730,132 · 1,766,645 · Down 36,513 · 1,884,077 · Down 8.2 percent

Work-sharing, short-time compensation · 14,208 · 14,751 · Down 543 · 25,912 · Down 45.2 percent

Former federal civilian employees · 5,727 · 5,691 · Up 36 · 8,211 · Down 30.3 percent

Newly discharged veterans · 4,902 · 4,814 · Up 88 · 4,667 · Up 5.0 percent

State additional benefits · 3,147 · 2,931 · Up 216 · 2,389 · Up 31.7 percent

Extended benefits · 13 · 30 · Down 17 · 82 · Down 84.1 percent

Total, all programs · 1,758,129 · 1,794,862 · Down 36,733 · 1,925,338 · Down 8.7 percent

Initial claims filed in federal programs, week ending August 29:

Program · Aug 29 · Aug 22 · Change · Year ago · Year-over-year

Former federal civilian employees · 388 · 337 · Up 51 · 527 · Down 26.4 percent

Newly discharged veterans · 409 · 413 · Down 4 · 317 · Up 29.0 percent

Two things in that table are worth more than their size. Work-sharing at 45.2 percent below last year is the strongest single argument in the release that employers are not preparing to cut — the gap has widened from 37 to 39.1 to 45.2 percent over three weeks. State additional benefits up 31.7 percent while the total fell 8.7 percent is the counterweight, and it counts people who have exhausted everything else. On 3,147 claims that is a signal you note and do not lean on.

No state was triggered on for the Extended Benefits program in the week ending August 22.

Impact on USD

• Slight bullish — a print a thousand above consensus, an insured-unemployment beat, and a four-week average that turned down together remove the labor-deterioration argument going into next week's meeting.
• The trend reversal matters more than the level: 207,500 to 206,000 on the four-week average ends the only labor series that had started rolling over.
• Continued claims beat by 6,000 and the prior week was revised down 4,000 — the measure of people who cannot find another job improved on both counts.
• No state posted a layoff increase above 1,000 for the third week in four, and Extended Benefits are triggered on nowhere in the country.
• Conditional and small. This shares its morning with producer prices and precedes consumer prices by a day; claims has not moved the dollar in months and will not this week.

Impact on US Indices (ES / NQ / YM)

• Slight bullish — layoffs remain rare, the trend improved, and a closed firing channel is what an equity market wants from labor data.
• Work-sharing 45.2 percent below last year says employers are not staging for cuts, which supports the margin story more than any single claims print.
• The warning sits in New York's comment: the layoffs it named were in health care, social assistance and educational services — the sector carrying essentially the entire payroll gain and the one the benchmark said was overstated by 96,000.
• Michigan up 2,149 and California up 2,221 in the advance week is a goods-and-technology combination worth watching, though unadjusted single-week state data revises hard.
• Freight names take a mixed read: transport prices up 13.0 percent over twelve months in this morning's producer report, but transportation and warehousing named among New York's layoffs.

Impact on Gold

• Slight bearish — a firm labor market with a falling four-week average weakens the case for the cuts gold has been pricing.
• Real yields are the transmission, and nothing here pushes them lower; the two-year still sits roughly 67 basis points above the funds-target midpoint.
• The offsetting bid is inflation, not employment — producer prices at 5.4 percent over twelve months this morning did more for gold than claims did against it.
• Forward conditional: a hot consumer print tomorrow restores the real-wage squeeze argument and this release is forgotten by the open.

What To Watch

• Consumer prices — Friday, 11 September, 8:30am ET. The last major data before the FOMC, sharing the morning with preliminary September Michigan sentiment. Everything this week defers to it.
• FOMC decision and dot plot — Tuesday and Wednesday, 15-16 September. Three dissents for a hike in July, a Chair who has retired forward guidance, and now a labor market that stopped deteriorating in the week before the meeting.
• Next claims — Thursday, 17 September, 8:30am ET, covering the week ending September 12. That is the reference week for the September payroll survey, so it carries more weight than a normal weekly print.
• Whether New York's attribution repeats. The spike itself unwound within a week, but a second state naming health care, social assistance or educational services would turn one comment into a pattern in the only sector still adding jobs.
• The four-week average against 199,000. One down week leaves it 3.5 percent above the August 1 low. Two more decides whether the August drift was a turn or noise.
• September employment situation — Friday, 2 October, 8:30am ET.

TLDR

Unemployment Insurance Weekly Claims, week ending 5 September 2026 (released 10 September, USDL 26-1503):

• Initial claims 206,000 (vs 205,000 est, 207,000 revised prior) — a thousand above consensus and effectively in line
• The 4-week average fell 1,500 to 206,000, its first decline in five weeks, ending the four-week climb off the August 1 low of 199,000
• Still 3.5 percent above that low, so the drift paused rather than reversed
• Insured unemployment 1,774,000 against 1,780,000 expected — a beat, with the prior week revised down 4,000; its 4-week average fell a third straight week
• Insured unemployment rate 1.2 percent adjusted and 1.1 percent unadjusted, both unchanged, against 1.3 and 1.2 a year ago
• Unadjusted claims rose 5,164 where factors expected 5,789 — a 625 gap, against 3,024 and 1,256 the two weeks before. The quietest mechanics in a month
• One state posted an increase above 1,000: New York, up 4,338, citing layoffs in transportation and warehousing, health care and social assistance, and educational services
• New York gave the whole spike back the following week, falling 3,571 to 14,666 — the largest state decline in the country
• Advance week increases came from California up 2,221 and Michigan up 2,149, the only two states above 1,000
• Work-sharing 14,208, down 45.2 percent year over year — employers are not staging for cuts, and the gap has widened three weeks running
• State additional benefits up 31.7 percent year over year while total claims fell 8.7 percent — tiny at 3,147, but it counts people who have exhausted everything else
• Veterans remain the exception: initial claims up 29.0 percent and continued claims up 5.0 percent year over year
• Four-week average 13.9 percent below last year; all-programs continued claims 8.7 percent below at 1,758,129

The one labor series that had started to turn stopped turning. The four-week average fell 1,500 to 206,000 after four straight increases, insured unemployment beat and was revised better, and the seasonal mechanics that drove the last two headlines went quiet — which makes last week's reading of a rolling layoff channel a week weaker than it looked, though at 3.5 percent above the August 1 low the drift has paused rather than reversed. The content is in the one state that cleared the reporting threshold. New York blamed transportation and warehousing, health care and social assistance, and educational services — the fastest-inflating sector in this morning's producer report at 13.0 percent, and the only sector holding up every payroll print in the calendar. That spike unwound entirely the following week, so it is an attribution to file rather than a trend to trade. Consumer prices tomorrow at 8:30am ET decide the week; the FOMC decides on 16 September.

_For informational purposes only. Not investment advice._