Jobless Claims Fall to 206K — But the Four-Week Average Crosses Back Above 200K

Fundamentals · 2026-08-20

Initial claims 206,000 for week ending August 15, down 6,000 from an upward-revised 212,000 (was 209,000); 4-week average 204,000, up 4,250 and back above 200K; continuing claims 1,799,000 for week ending August 8, up 18,000 and above the ~1.790M consensus; insured unemployment rate 1.2 percent, unchanged; unadjusted initial claims 172,080 versus 194,217 in the comparable week of 2025, down 11.4 percent year over year; largest state increase Michigan +1,931; no state triggered on Extended Benefits.

What Is This?

• What it is: The Department of Labor's weekly count of Americans filing for unemployment insurance, compiled by the Employment and Training Administration from state agency reports and released every Thursday at 8:30am ET. Initial claims count new filings; continuing claims (insured unemployment) count people still drawing benefits, and lag initial claims by one week. Both are published seasonally adjusted and raw.
• Why it matters: It is the highest-frequency labor reading that exists, and it is the series that would show a cracking labor market first. With the Fed holding at 3.50-3.75 percent and three members dissenting for a hike at the July 29 meeting, the dovish case depends on labor deterioration that this series is not yet delivering.
• How to read it: The weekly print is noise; the four-week moving average is the signal. Initial claims measure firing, continuing claims measure how hard it is to get rehired — the two moving in opposite directions is the statistical fingerprint of the "low-fire, low-hire" market. Also treat every headline as provisional: the prior week is revised almost every single week, and this week it was revised up by 3,000.

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 August 15, below the roughly 210,000 consensus and down 6,000 on the week. The beat is softer than it looks. The previous week was revised up by 3,000, from 209,000 to 212,000, so the decline is measured from a worse starting point than the one originally reported — the same upward-revision pattern that has been quietly flattering these headlines for weeks.

The four-week moving average is the number that actually moved. It rose 4,250 to 204,000, crossing back above the 200,000 line, and the prior average was itself revised up 750 to 199,750. So the smoothed trend in firings is drifting higher even as the weekly print falls. That is not a deteriorating labor market, but it is no longer an improving one.

Continuing claims tell the other half of the story. Insured unemployment rose 18,000 to 1,799,000 for the week ending August 8, above the roughly 1.790 million expected, with the prior week revised up 4,000 to 1,781,000. The four-week average climbed to 1,789,000. This still sits below the 1.814 million cycle high, so the trend has not broken out — but people who lose jobs are taking longer to find new ones, which is the textbook "low-hire" condition. The insured unemployment rate held at 1.2 percent.

The seasonal mechanics explain why the two series diverged, and they are worth understanding rather than skipping. Unadjusted initial claims fell 17,123 to 172,080, a 9.1 percent drop when the seasonal factors had only expected 6.4 percent — firing came in better than the calendar implied, which pushed the adjusted headline down. Unadjusted continuing claims fell 17,509, or 1.0 percent, when the factors expected a 2.0 percent decline — roughly 18,000 fewer people left the benefit rolls than the calendar implied, which is precisely the 18,000 increase that showed up in the adjusted figure. Neither number is an accident of adjustment. They are measuring two genuinely different things.

The year-over-year comparison is the most under-reported part of this release and cuts firmly against the labor-cracking thesis. Unadjusted initial claims of 172,080 compare with 194,217 in the same week of 2025 — down 11.4 percent. Unadjusted continuing claims of 1,797,050 compare with 1,955,067 a year ago, down 8.1 percent, with the unadjusted rate at 1.2 percent versus 1.3 percent. Total continued weeks claimed across all programs came in at 1,839,126 for the week ending August 1, against 2,005,772 in the comparable week last year, down 8.3 percent. Fewer people are filing and fewer people are collecting than at this point in 2025, on every measure.

Set against the running narrative, this lands the same way the Philadelphia Fed's 47.4 print did an hour earlier on the same morning: it removes the growth excuse. A Fed with inflation near 3.4-3.5 percent, three voters already dissenting for a hike, and roughly zero cuts priced for 2026 now has a claims series running 11 percent below last year and an insured unemployment rate of 1.2 percent. The rising four-week average and the sticky continuing-claims trend are the only threads the doves have left, and neither is yet strong enough to pull.

The Internals

Seasonally adjusted:

Metric · Current · Prior (revised) · Change · Read

Initial claims, week ending Aug 15 · 206,000 · 212,000 · -6,000 · Beat the ~210,000 consensus

Initial claims 4-week average · 204,000 · 199,750 · +4,250 · Back above 200,000

Continuing claims, week ending Aug 8 · 1,799,000 · 1,781,000 · +18,000 · Above the ~1.790M consensus

Continuing claims 4-week average · 1,789,000 · 1,786,500 · +2,500 · Grinding higher, below cycle high

Insured unemployment rate · 1.2 percent · 1.2 percent · unchanged · Historically very low

Revisions to the prior week, all upward:

Series · Originally reported · Revised to · Revision

Initial claims · 209,000 · 212,000 · +3,000

Initial claims 4-week average · 199,000 · 199,750 · +750

Continuing claims · 1,777,000 · 1,781,000 · +4,000

Continuing claims 4-week average · 1,785,500 · 1,786,500 · +1,000

Unadjusted, versus what the seasonal factors expected:

Series · Actual change · Seasonals expected · Result

Initial claims, 172,080 · -17,123 or -9.1 percent · -12,077 or -6.4 percent · Better than expected, pushed adjusted headline down

Continuing claims, 1,797,050 · -17,509 or -1.0 percent · -35,615 or -2.0 percent · Worse than expected, pushed adjusted level up

Year-over-year, unadjusted:

Series · This week 2026 · Comparable week 2025 · Change

Initial claims · 172,080 · 194,217 · -11.4 percent

Continuing claims · 1,797,050 · 1,955,067 · -8.1 percent

Continued weeks, all programs · 1,839,126 · 2,005,772 · -8.3 percent

Unadjusted insured unemployment rate · 1.2 percent · 1.3 percent · -0.1pp

State And Program Detail

Largest state moves in initial claims, week ending August 8:

State · Change · Note

Michigan · +1,931 · Largest increase, consistent with auto retooling shutdowns

New York · +1,379 · Second largest increase

Texas · +1,324 · Third largest increase

South Carolina · +1,268 · Notable for the size of the state

Illinois · +994 · Fifth largest increase

Ohio · -252 · Largest decrease

Iowa · -103 · Second largest decrease

Kentucky · -87 · Third largest decrease

Louisiana · -41 · Fourth largest decrease

North Dakota · -33 · Fifth largest decrease

The asymmetry matters: the five largest increases total roughly 6,900 filings while the five largest decreases total just 516. Upward pressure was concentrated rather than broad — Michigan alone accounts for more than a quarter of it — and no state saw filings fall meaningfully. A national headline that improved on the week did so despite its biggest movers, not because of them.

Highest insured unemployment rates, week ending August 1: New Jersey 2.6, Puerto Rico 2.6, Rhode Island 2.2, Massachusetts 2.1, Minnesota 2.1, Oregon 2.0, California 1.9, Washington 1.9, Connecticut 1.7, Nevada 1.7, New York 1.7, Pennsylvania 1.7.

Federal programs, and a genuinely quiet corner of the report:

Program · Level · Change

Initial claims, former Federal civilian employees, week ending Aug 8 · 449 · +48

Initial claims, newly discharged veterans, week ending Aug 8 · 489 · +81

Continued weeks, former Federal civilian employees, week ending Aug 1 · 6,015 · -490

Continued weeks, newly discharged veterans, week ending Aug 1 · 4,610 · -345

No state was triggered on the Extended Benefits program during the week ending August 1 — the cleanest single indicator that no state-level labor market has deteriorated far enough to hit its statutory trigger.

Impact on USD

• Mixed, lean bullish — a sub-consensus headline with the insured unemployment rate at 1.2 percent gives the Fed no labor-market reason to ease.
• Claims running 11.4 percent below the comparable 2025 week undercuts the labor-deterioration case the doves need.
• Partial offset: continuing claims at 1,799,000 came in above consensus and the four-week average crossed back above 200,000, so this is not a one-way print.
• Same-morning confirmation from Philly Fed at 47.4 compounds the effect — two prints, one direction, no growth excuse.
• Scale caveat: a 4,000 beat on weekly claims is not on its own a dollar driver; it moves the dollar by nudging September odds.

Impact on US Indices (ES / NQ / YM)

• Mixed, lean bearish — there is no earnings channel here, only a rates channel, and a tight labor market is a hawkish input.
• With roughly zero cuts priced for 2026 and hike odds live into year-end, low claims read as policy risk rather than growth comfort.
• NQ carries the most duration risk of the three and is the most exposed to a hawkish repricing into the dot plot.
• Consumer-facing names take the opposite read — 1.2 percent insured unemployment supports spending into the back half of the year.
• Watch the four-week average rather than the weekly headline; a sustained move above 210,000 is what would flip this back to a growth-scare trade.

Impact on Gold

• Slight bearish — firm labor data supports real yields and delays the cut that the metal wants.
• No labor deterioration means no near-term policy easing, removing the most direct bullish catalyst.
• Offsetting bid: sticky inflation near 3.4-3.5 percent and a Fed with live hike dissent keeps the stagflation hedge in play.
• Claims are a second-order driver for gold at these levels — CPI and the dot plot dominate.
• Conditional: a soft August CPI on September 11 matters far more to the level than any single claims print.

What To Watch

• Next claims report — Thursday, August 27. Watch whether the four-week average holds above 200,000 and whether this week's 206,000 gets revised up, as the prior four prints have been.
• August CPI — Thursday, September 11. The dominant input to the September decision; strong labor plus hot inflation is the combination that empowers the hike dissenters.
• FOMC decision and dot plot — September 15-16. Fifth consecutive hold was 9-3 with three members preferring a 25bp hike; claims this low remove a dovish argument.
• Continuing claims versus the 1.814 million cycle high. A decisive break above it would be the first hard evidence that low-hire is turning into something worse.
• Michigan in coming weeks. A +1,931 spike is usually auto retooling and reverses within two to three weeks; if it does not reverse, it is a genuine manufacturing layoff signal.

TLDR

Unemployment Insurance Weekly Claims (week ending August 15, released August 20):

• Initial claims: 206,000 (vs ~210,000 est, 212,000 prior revised) — beat, but off an upward-revised base
• Prior week revised up 3,000, from 209,000 to 212,000 — the recurring pattern
• 4-week moving average: 204,000 (from 199,750) — up 4,250 and back above 200,000
• Continuing claims: 1,799,000 (vs ~1,790,000 est, 1,781,000 prior) — up 18,000, a miss on the high side
• Continuing claims 4-week average: 1,789,000 — still below the 1.814 million cycle high
• Insured unemployment rate: 1.2 percent — unchanged, versus 1.3 percent a year ago
• Unadjusted initial claims: 172,080 vs 194,217 a year ago — down 11.4 percent
• All-programs continued weeks: 1,839,126 vs 2,005,772 a year ago — down 8.3 percent
• Largest state increases: Michigan +1,931, New York +1,379, Texas +1,324, South Carolina +1,268
• No state triggered on Extended Benefits

Firing stays historically low and running 11 percent below last year, while rehiring stays slow and the four-week average creeps back over 200,000 — low-fire, low-hire, unbroken. Landing the same morning as a five-year high in Philly Fed manufacturing, this leaves the Fed with no labor-market excuse to ease and hands the three hike dissenters another data point. Watch next Thursday's revision on August 27, August CPI on September 11, and the dot plot on September 15-16.

_For informational purposes only. Not investment advice._