Those unemployment numbers

The weekly unemployment claims were very positive today.

(Bloomberg) -- The number of Americans filing claims for unemployment benefits fell last week to the lowest level since September 2008 as the economic recovery encourages companies to fire fewer workers.
Initial jobless claims declined to 466,000 in the week ended Nov. 21 from 501,000 a week earlier, Labor Department figures showed today in Washington. The number of people collecting unemployment insurance dropped in the prior week, while those getting extended payments also declined.

Great news all around, right? Wrong.
The non-seasonally adjusted numbers tell a far different story.

Weekly Initial Claims (SA): -35,000
Weekly Initial Claims (NSA): +68,080

Continuing Ins. Unemployment (SA): -190,000
Continuing Ins. Unemployment (NSA): +187,642

Some of the decline in claims last week may have resulted from the way the government adjusts the data to smooth out seasonal fluctuations, economists said.

No, ALL of the decline, and more, resulted from the way the government adjusts the data to smooth out seasonal fluctuations.

The non-seasonally adjusted numbers are the real numbers, where one unemployment claim equals one person on unemployment.
As for people on the federal government's extended benefits:

EUC 2008: +16,370

Subject Meta: 

Forum Categories: 


I was writing up weekly initial claims and saw a huge pattern. Every week the BLS would claim "big drop" but they would revise the previous week significantly upward so it looked like a "big drop" every week, when without revisions it would have looked like it was increasing to flat.

And as we could see when they finally had to come sort of clean on the October report, nope, unemployment is rising.

I do go with the seasonal, but we have noticed multiple divergence from the NSA to SA this year. It should all iron out in a year average but there are a lot of things happening this year which are not normal.

They have a huge regression analysis algorithm for I went looking into what are the current SA assumptions and I could not find them, just the monthly elements listed.

That's strange for these should at least give an element window into what is in their seasonal adjustments algorithms.

I imagine it's something about seasonal summer help but it should be temporary hiring for the holidays.

Anyway, bottom line is these weekly initial claims have turned into headline cheerleading and just have way too much noise in them to tell me much.

I think the entire BLS data collection needs to be ravamped. It's 2009 and it just doesn't give an accurate snapshot, esp. these weeklies, on what is going on in the U.S. labor market.