- by New Deal democrat
The BLS reported that for the week ending Nov. 21, seasonally adjusted initial jobless claims were 466,000. Last week's number was revised to 501,000. This is the best showing since "Black September" 2008 when the economy nearly ground to a panicked halt.
The 4-week moving average was 496,500, a decrease of 16,500 from the previous week's revised average of 513,000. The 4 week seasonally adjusted moving average is now about 24% lower than the peak of 658,750 on April 3 of this year.
Unadjusted, there were 543,926 new claims, an increase of 68,080 from the week before, and well below the 609,138 initial claims in the same week last year. In unadjusted terms, this was the best new claims number, relative to normal seasonal adjustment, in well over a year.
Because the BLS normally surveys business payrolls in the week ending the 12th of the month, this decrease if it persists won't show up until the December jobs number. If it does, according to my previous research, this indicates that jobs are actually being added to the economy. In this position I am at odds with people like Berkeley Economics Professor Brad DeLong and Calculated Risk, who say that the claims number must drop ot 400,000 before jobs are added. A number like today's is why I said I have no problem being proven wrong, provided that it is done quickly!
In that regard, here is a repost of some numbers I posted two weeks ago:
At the time of the 501,250 4 week average of new jobless claims reading in 1990, payrolls lost 160,000 that month and 211,000 the next. In 2001, the new jobless claims high of 489,250 coincided with payroll losses of 325,000 that month and 292,000 the next. This year, we have already seen in August new jobless claims in the 560,000-570,000 range coinciding with a payroll loss of 151,000.
The Joy of Term Papers
6 minutes ago
5 comments:
The non seasonally adjusted number came in lower than I expected the implication is that there hasn't been the season hiring that the numbers expected the adjustment factor two weeks from is a monstrous 1.40 if they don't get the non seasonally adjusted number they were expecting ( which is looking increasingly likely) you are going to see continued jobless claims drop rather dramatically,,,,
Seasonal adjustment is a big factor in these numbers. Using the previous week's seasonal adjustment factor would have yielded SA first time claims of 572,500. Given the huge weekly variations in firings at this time of year, and the possibility that this year's situation is atypical, we need more than one week of data to get optimistic.
Is there ever a time to seriously question a seasonal adjustment? I mean, have there been times when the consensus was that SA was radically off in either direction? Or are the SAs generally accepted as reliable?
PJ:
I disagree with your comment for several reasons.
1. I am in no way relying on just one week's data. In fact, here's the last 15 weeks:
2009-08-15 580000
2009-08-22 574000
2009-08-29 576000
2009-09-05 557000
2009-09-12 550000
2009-09-19 534000
2009-09-26 554000
2009-10-03 524000
2009-10-10 520000
2009-10-17 531000
2009-10-24 532000
2009-10-31 514000
2009-11-07 505000
2009-11-14 501000
2009-11-21 466000
To the contrary, in the face of that steady decline, in my opinion it is cherry-picking NOT to be optimistic.
2. It is also cherry picking not to use this week's seasonal adjustment, when the adjustment is that in a typical year, layoffs increase rapidly from late September to the end of December.
For example, in the last 2 non-recession years, from Labor Day until this week NSA new jobless claims had increased ~80,000. Last year, in the same time period they increased about 220,000. This year, they have increase ~130,000 -- much closer to their typical pattern.
3. I was very clear in the post itself to say "this decrease if it persists won't show up until the December jobs number". What part of "if it persists" suggests that I am relying on one week's number?
NDD - I agree that the numbers have shown steady improvement. I didn't say that using last weeks adjustment factor was more appropriate than using this week's. My point is just that these are noisy data, and the noisiest part is the seasonal adjustment factor.
We can combine the continuing claims, exhaustion, and first time claims data to get a measure of hiring off the unemployment rolls, and until recently hiring was declining at roughly the same rate as firing (measured by first times claims). Only in the last month has hiring picked up slightly, but it is still at only 60% of normal levels. So we can get a lower rate of firing and lower first time claims, but still be in a deteriorating employment environment.
Finally, the seasonal adjustment methods make exaggerated adjustments when claims are high. All through the summer and early fall the seasonal adjustments were making first time claims look worse than they really were; now we are into the heavy layoff period of the year and seasonal adjustment is making the numbers appear better than they really are. In fact the big reduction in firings was over the summer, but the employment situation stagnated for several months, and has improved slightly again in the last month, but is still poor.
As brodero notes, the exaggerated seasonal adjustments will make the first time claims data look very good for the next few months. This will get hopes up for good monthly employment reports -- hopes that will be disappointed.
Like your blog very much. But you guys seem a bit touchy, as if you think everybody is hostile. Not so. I'm just searching for the truth in an economy that's difficult to analyze.
Post a Comment