MaxedOutMama
Thursday, July 16, 2009
Weekly Claims Problems
The seasonal adjustments in weekly claims are always a bit iffy at certain points, and this is one of them.
The following are some numbers from the weekly claims from the last few years (which you can access here). The number in the first row in red is the NSA weekly initial claims, and the second number in blue is NSA continuing claims. The number in black between those two is the seasonal factor:
So the NSA claims for the week of 6/16/07 are 89% of the SA claims.
For 2009, the NSA figures (some of which are still preliminary) for the same period go:
6/13/2009 558,407/ 6,113,273
6/20/2009 568,552/ 6,078,496
6/27/2009 559,857/ 6,071,352
7/04/2009 581,145/ 6,135,066
7/11/2009 667,534/
We don't have the continuing claims number for this week because that is delayed a week. Seasonal factors are published ahead of the figures, and are as follows:
The Dept of Labor people who issue this report do a very fine job, and they note that a shift in the timing of auto layoffs might be distorting the relationship of SA and NSA claims and continuing claims. Bloomberg article:
So we could be seeing a knock-on effect from the depressed manufacturing areas. It's hard to know. It's also possible that the auto parts dealers followed the regular shutdown schedule. My impression based on some of the county data is that we are moving into the scorched-earth cycle for the more depressed areas in which small businesses start shedding employees or shutting down, and small retail mostly gives up the ghost.
I hope I'm wrong, but I have unemployment heading to around 11% in 2010.
There is another problem with seasonal adjustments in times of rapid economic change. Conceptually, closing of school systems or the closing of auto plants has a set effect on jobs related to the actual number of those jobs, and if claims are high from other layoffs, the Seasonal Factor (because it is expressed as a percent of the whole) may either under or over correct. This problem is kind of blatantly obvious in continuing claims this week, which are reported to have dropped 642,000, when in actuality they rose close to 64,000.
If you look at actual claims and continuing claims, there appears to be a flattening effect over the last six weeks. We may, however, be seeing a surge in ancillary service small business shutdowns in the most depressed areas, and if we are, we will see some very ugly numbers in August and September, although not necessarily in the weekly claims reports. The high number of contractor/individual proprietorships relative to corporate employment in those areas often distorts the claims reports versus unemployment surveys.
The following are some numbers from the weekly claims from the last few years (which you can access here). The number in the first row in red is the NSA weekly initial claims, and the second number in blue is NSA continuing claims. The number in black between those two is the seasonal factor:
| 06/16/2007 | 290,951 | 89.0 | 327,000 | 2,278,797 | 90.8 | 2,510,000 | 1.7 | 1.9 | 131,660,675 |
|---|---|---|---|---|---|---|---|---|---|
| 06/23/2007 | 292,583 | 91.9 | 318,000 | 2,353,893 | 91.3 | 2,578,000 | 1.8 | 2.0 | 131,660,675 |
| 06/30/2007 | 300,348 | 93.0 | 323,000 | 2,303,357 | 89.9 | 2,562,000 | 1.7 | 1.9 | 131,660,675 |
| 07/07/2007 | 417,554 | 131.4 | 318,000 | 2,581,098 | 99.3 | 2,599,000 | 2.0 | 2.0 | 132,170,980 |
| 07/14/2007 | 383,839 | 121.5 | 316,000 | 2,588,709 | 100.1 | 2,586,000 | 2.0 | 2.0 | 132,170,980 |
| 07/21/2007 | 298,366 | 95.7 | 312,000 | 2,463,232 | 96.4 | 2,555,000 | 1.9 | 1.9 | 132,170,980 |
| 07/28/2007 | 257,426 | 83.7 | 308,000 | 2,445,524 | 94.9 | 2,577,000 | 1.9 | 1.9 | 132,170,980 |
| 06/14/2008 | 349,255 | 89.6 | 390,000 | 2,852,599 | 90.6 | 3,149,000 | 2.1 | 2.4 | 133,382,559 |
|---|---|---|---|---|---|---|---|---|---|
| 06/21/2008 | 358,159 | 91.3 | 392,000 | 2,855,557 | 90.6 | 3,152,000 | 2.1 | 2.4 | 133,382,559 |
| 06/28/2008 | 368,544 | 91.0 | 405,000 | 2,857,438 | 88.8 | 3,218,000 | 2.1 | 2.4 | 133,382,559 |
| 07/05/2008 | 401,672 | 109.3 | 367,000 | 3,118,724 | 98.4 | 3,169,000 | 2.3 | 2.4 | 133,690,617 |
| 07/12/2008 | 483,981 | 125.2 | 387,000 | 3,164,970 | 100.0 | 3,165,000 | 2.4 | 2.4 | 133,690,617 |
| 07/19/2008 | 411,408 | 99.7 | 413,000 | 3,208,848 | 97.0 | 3,308,000 | 2.4 | 2.5 | 133,690,617 |
| 07/26/2008 | 376,123 | 85.1 | 442,000 | 3,166,763 | 94.7 | 3,344,000 | 2.4 | 2.5 | 133,690,617 |
For 2009, the NSA figures (some of which are still preliminary) for the same period go:
6/13/2009 558,407/ 6,113,273
6/20/2009 568,552/ 6,078,496
6/27/2009 559,857/ 6,071,352
7/04/2009 581,145/ 6,135,066
7/11/2009 667,534/
We don't have the continuing claims number for this week because that is delayed a week. Seasonal factors are published ahead of the figures, and are as follows:
| 06/13/2009 | 558,407 | 91.2 | 612,000 | 6,113,273 | 90.5 | 6,755,000 | 4.6 | 5.1 | 133,683,433 |
|---|---|---|---|---|---|---|---|---|---|
| 06/20/2009 | 568,552 | 90.3 | 630,000 | 6,078,496 | 90.4 | 6,724,000 | 4.5 | 5.0 | 133,683,433 |
| 06/27/2009 | 90.7 | 87.8 | |||||||
| 07/04/2009 | 102.2 | 97.8 | |||||||
| 07/11/2009 | 128.0 | 100.1 | |||||||
| 07/18/2009 | 104.8 | 97.8 | |||||||
| 07/25/2009 | 86.9 | 94.9 |
The Dept of Labor people who issue this report do a very fine job, and they note that a shift in the timing of auto layoffs might be distorting the relationship of SA and NSA claims and continuing claims. Bloomberg article:
Initial jobless claims dropped by 47,000 to 522,000, lower than forecast, in the week ended July 11, from a revised 569,000 the prior week, the Labor Department said today in Washington. The number of people collecting unemployment insurance plunged by a record 642,000, also reflecting seasonal issues surrounding the closures at carmakers.I am not sure that there is much of a distortion from the timing of the auto layoffs, because when I look at the shape of these curves they look somewhat comparable to prior years, and when I look at the states that showed early pops in July UI, they are the manufacturing states:
A Labor analyst said the distortions may play havoc with claims data for another couple of weeks. General Motors Co. and Chrysler Group LLC accelerated shutdowns this year heading into bankruptcy, months before the traditional July closings. Through the gyrations, job losses may subside amid signs the housing and manufacturing slumps are easing.
| State | Change | State Supplied Comment | |
| MN | +1,176 | No comment. | |
| KY | +1,203 | No comment. | |
| FL | +1,679 | Layoffs in the construction, trade, service, and manufacturing industries, and agriculture. | |
| IA | +3,351 | Increase due to temporary holiday layoffs. | |
| OH | +4,240 | Layoffs in the service and manufacturing industries. | |
| IN | +5,430 | Layoffs in the automobile, trade, service, and manufacturing industries. | |
| WI | +5,838 | Layoffs in the manufacturing industry. | |
| NY | +8,913 | Layoffs in the transportation and service industries. | |
| MI | +12,144 | Most industries posted increases. |
So we could be seeing a knock-on effect from the depressed manufacturing areas. It's hard to know. It's also possible that the auto parts dealers followed the regular shutdown schedule. My impression based on some of the county data is that we are moving into the scorched-earth cycle for the more depressed areas in which small businesses start shedding employees or shutting down, and small retail mostly gives up the ghost.
I hope I'm wrong, but I have unemployment heading to around 11% in 2010.
There is another problem with seasonal adjustments in times of rapid economic change. Conceptually, closing of school systems or the closing of auto plants has a set effect on jobs related to the actual number of those jobs, and if claims are high from other layoffs, the Seasonal Factor (because it is expressed as a percent of the whole) may either under or over correct. This problem is kind of blatantly obvious in continuing claims this week, which are reported to have dropped 642,000, when in actuality they rose close to 64,000.
If you look at actual claims and continuing claims, there appears to be a flattening effect over the last six weeks. We may, however, be seeing a surge in ancillary service small business shutdowns in the most depressed areas, and if we are, we will see some very ugly numbers in August and September, although not necessarily in the weekly claims reports. The high number of contractor/individual proprietorships relative to corporate employment in those areas often distorts the claims reports versus unemployment surveys.
Comments:
<< Home
I have such respect for you as a scientist. I took one look at the number, figured it was f*cked up and went on with my day.
When's your birthday? I want to send you another First Edition.
When's your birthday? I want to send you another First Edition.
Thanks for the work.
I hope unemployment doesn't drop any lower than 11%. That would be quite the improvement over my expectations.
Post a Comment
I hope unemployment doesn't drop any lower than 11%. That would be quite the improvement over my expectations.
<< Home
