About the decline in jobless claims this week- Lying in plain site

[AllentownJake]

This is what the media reports

In the week ending Nov. 21, the advance figure for seasonally adjusted initial claims was 466,000, a decrease of 35,000 from the previous week's revised figure of 501,000. The 4-week moving average was 496,500, a decrease of 16,500 from the previous week's revised average of 513,000.

However, this is what you read if you actually read the report

The advance number of actual initial claims under state programs, unadjusted, totaled 543,926 in the week ending Nov. 21, an increase of 68,080 from the previous week. There were 609,138 initial claims in the comparable week in 2008.

Honestly, 80,000 jobs are a seasonal adjustment? You've got to be fucking kidding me.


http://www.dol.gov/opa/media/press/eta/ui/current.htm

[Greyhound]

1. It's the only way to make this pig appealing.

Elimination of any static baseline is the only way to conceal what nearly everyone already knows. Government numbers are meaningless anymore, Wall Street doesn't use them except to run the short-term pump-and-dump scams.
:kick: & R

[AllentownJake]

2. The funny thing is they report the truth in plain sight

They just give the media what they want reported in the first paragraph. The rest of the document than tells the truth.

[FBaggins]

3. There's nothing at all unusual about that.

There really is that much seasonal variation (more at times).

They gave you the relevant numbers for comparison as well... You can see how high the unadjusted number was last year. It isn't reasonable to call it a "lie"

if you graph the unadjusted data for multiple years you can see that it's a classic case for a seasonal adjustment.

[AllentownJake]

4. 18 fucking percent?

The actual number went up 68,000 and they are calling it a decrease.

[FBaggins]

7. Absolutely

If it were a small variation then there wouldn't be much point in doing an adjustment, would there?

There are simply times of the year when companies traditionally hire/fire more people. You don't want that variation to give people the wrong impression. This report is a good example. Reporting just raw unadjusted figures would make it look like things were getting worse when they were actually improving.

[AllentownJake]

8. Can you show me an instance

Where they seasonally adjusted job loss numbers that were lower, higher.

[tritsofme]

9. Look at the payroll numbers over the past year

The top being total seasonally-adjusted non-farm employment, and the bottom being the non-seasonally adjusted data:

Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct
2009 134333 133652 133000 132481 132178 131715 131411 131257 131038(P) 130848(P
2009 132302 132138 132077 132336 132720 132625 131127 131046 131399(P) 132040(P)

(data from here http://data.bls.gov/cgi-bin/surveymost?ce)

Seasonal adjustments allow us to observe events at a constant from month to month, no conspiracy needed.

[AllentownJake]

10. I read up on the assumptions for Seasonal Adjustment this time period

1) Construction
2) Holiday Help

Both assumptions are off for this year, for one thing there isn't as many people doing construction because housing starts and commercial real estate starts took a dive in September.

Holiday Help, not hiring this year as much in previous years passed.

If they used those two assumptions in this data set. They are fucking wrong.

[FBaggins]

13. It isn't that simple

This is the number of initial layoffs... so the amount of HIRING this year for the holidays doesn't have much to do with that. Construction also isn't a good reply because last year at this time was well into the crash

Moreover... starts and CRE are ALSO seasonally adjusted numbers. :)

I've been on my iphone which makes it harder (and my posts shorter)... but I should be able to find you a good example shortly.

[AllentownJake]

14. Ok, respect your opinion

I'm looking at the layoffs in November and October on http://www.dailyjobcuts.com/

Heavy in Media, Municipal/State government, Education, and Manufacturing.

[FBaggins]

19. Sure.

I took a look at the last few years of the data (link below).

The first report of 2006 saw an non-seasonally adjusted figure of 555,114 - (which would appear to be pretty bad)... but a seasonally adjusted figure of 318,000 (which is pretty good but not spectacular). The non-adjusted number was a full 75% higher.

But reporting the NON-adjusted number as the headline would be the LESS honest option, because it ignores the fact that holiday season hires tend to get laid off at that point... it isn't a sign of massive employment damage.

Take a look just a month or so later and the non-adjusted figure is cut more than in half (269,571)... which was pretty close to the seasonally adjusted figure (286,000). That September the NSA figure is relatively unchanged (240,231) but the reported SA figure was about 30% higher(315,000).


http://www.ows.doleta.gov/unemploy/wkclaims/report.asp

Updated - better link that allows you to pick your own dates:

http://www.ows.doleta.gov/unemploy/claims.asp

[AllentownJake]

21. Question

In January of this year, there is less seasonally adjusted hiring going on in the retail sector and areas that serve the retail sector, are they going to use the same seasonal adjustment calculation they used in 2008? Because that would be kind of silly, if you make the same assumption about what is going on make a statistical adjustment when the fundamental to the adjustment didn't really happen.

Of course the stores could be letting go of full-time staff and that would be buried in such a report.

Which is what I'm saying about this report, did they make the same assumptions from 2008, when the world was a lot different.

[FBaggins]

24. There is an adjustment to the adjustment... but

there's no question that the seasonal adjustment can't account for non-seasonal factors... and that's what we dig for.

The seasonal adjustments are set based on movement over a number of prior years... so unusual years tend to average out. They get re-benchmarked on a periodic basis.

The lower hiring expected for the holiday season may very well result in lower initial claims figures in January... because some of the people who would have been laid off haven't been hired... and thus won't be laid off. The same issue will show up the other way in the jobs numbers because the SA will assume that people are being hired seasonally and adjust the monthly number downward... when people aren't actually seeing as much seasonal hiring. So the "core" number (to mis-use a label) of new hires to "real" lasting jobs will likely be higher than reported.

There are other things that impact these figures. The auto plants shut down for a couple/few weeks each year for retooling and tens of thousands of auto workers are technically "laid off" and then rehired (I think it's really just a gimmick the unions negotiated with the government so that they could get unemployment benefits for a couple weeks even though they aren't unemployed). Something similar happens in some states when teachers are "laid off" for the summer.

When the SA misjudges the week the auto layoff will occur, you'll see a really odd number one week and then an equally odd number in the other direction the next. An unsophisticated glimpse at the data would make it seem like things got really bad and then really improved (or visa-versa) when reality was just a mis-timed adjustment.

That's why reading the whole report for the larger picture is so usefull.

[AllentownJake]

25. I read the whole report

I think they are making some silly assumptions. Like I said, the reading I'm doing on-line is showing a large number of municipal/state, manufacturing, and media lay-offs being announced the past two months.

:shrug:

[FBaggins]

26. Ok... I think the main point from TFM and my posts

was that it's unreasonable to call it "lying" when both the raw figures and the adjusted ones are reported and the prose of the report gives the rest of the data.

The "headline" number accurately describes the employment/layoff situation. Still weak but improving significantly over just a few months ago. It's clearly a "better than expected" figure.

[tritsofme]

5. So now you are taking issue with seasonal adjustments?

The use of seasonal adjustments is a long established method in statistics.

[AllentownJake]

6. The actual claims went up 68,000

For fuck's sake seasonal adjustment my ass.

[branders seine]

11. you know what a professional turd polisher is?

Shitty.


If the government didn't lie to us, there would be armed rebellion in the streets.

[neverforget]

12. Up is the new down. Clap harder.

[TreasonousBastard]

15. Did you even bother to read the bottom of the page you linked...

that part where they broke down the states with reduced layoffs?

And, you do realize that the seasonal adjustment deals with jobs like bait shops and marinas in New England, farming and produce operations in Minnesota, construction in Wisconsin... Jobs that typically disappear as winter comes upon us.

[AllentownJake]

17. I read it

Are those seasonably adjusted as well...btw they were siting hiring in those states the negative number meant negative claims.

"Fewer layoffs in the trade, service, and manufacturing industries." Does not strike me as an explanation one would give for higher claims.

What were they explaining 2008 numbers, seasonably adjusted numbers or last week's numbers. If it was last week's number how could there be fewer claims when there were actually 68,000 more claims?

[TreasonousBastard]

22. It clearly states that they are not seasonally adjusted, but...

I just realized that it's for the prior week. The seasonal layoffs started to kick in for real in the last week.

Nonetheless, the comparison with November of last year shows improvement.

[AllentownJake]

23. Really a year after the Treasury Secretary threatened congress with Martial Law

Is a more stable time than a month after. Who would have thought that.

[CreekDog]

16. I just wish you would stick with the objectionable stuff when you complain

you are frequently right, but in cases like this, you sometimes see conspiracies where they just aren't.

[AllentownJake]

18. There were 68,000 actual more claims

and the data reported to the media was statistically adjusted to show that there were less claims than last week.

[dysfunctional press]

20. all of the numbers that we're hearing will be revised after december 25th.

it's all a RUSE to get people shopping, in the insane hope that a 'recovery' can be a self-fulfilling prophecy.