"Economists are pessimists: they've predicted 8 of the last 3 depressions."
--Barry Asmus

The Required Disclosures

The information presented in this blog and its individual articles is provided for informational use only and should not be considered investment advice or an offer for a particular security. The contents reflect the views and opinions of the individual writer as of the date the article was written and do not necessarily represent the views of the individual writer on the current date. They also do not in any way, shape, or form represent the views of the Firm Never-To-Be-Named. Any such views are subject to change at any time based upon market or other conditions and The Great Redoubt and its individual writers disclaim any responsibility to update such views. These views should not be relied on as investment advice, and because investment decisions for any security are based on numerous factors, may not be relied on as an indication of trading intent on behalf of any contributor to The Great Redoubt. Neither The Great Redoubt nor any individual author can be held responsible for any direct or incidental loss incurred by applying any of the information offered. Please consult your tax or financial advisor for additional information concerning your specific situation.

Thursday, January 6, 2011

Seasonal Adjustment

Something interesting to keep in mind about the 409,000 new first time jobless claims, is that that figure is the seasonally adjusted figure. The actual, unadjusted figure for first time jobless claims was 577,279 for the same period, up 52,038 from the week ending Christmas Day. Quite a substantial difference from 409,000 (up 21,000), hey?
So what is this "seasonal adjustment" anyway? The US Census Bureau has a pretty good FAQ on Seasonal Adjustment, and here's the explanation they give: "Seasonal adjustment is the process of estimating and removing seasonal effects from a time series in order to better reveal certain non-seasonal features. Examples of seasonal effects include a July drop in automobile production as factories retool for new models and increases in heating oil production during September in anticipation of the winter heating season."
As to why seasonal adjustment is used, the explanation is that "seasonal movements are often large enough that they mask other characteristics of the data that are of interest to analysts of current economic trends. For example, if a month has a different seasonal tendency towards high or low values it can be difficult to detect the general direction of a time series' recent monthly movement (increase, decrease, turning point, no change, consistency with another economic indicator, etc). Seasonal adjustment produces data in which the values of neighboring months are usually easier to compare. Many data users prefer seasonally adjusted data because they want to see those characteristics that seasonal movements tend to mask, especially changes in the direction of the series."
And finally, here's what they have to say about what what sort of seasonal effects are removed: "Seasonal adjustment procedures for monthly time series estimate effects that occur in the same calendar month with similar magnitude and direction from year to year. In series whose seasonal effects come primarily from weather (rather than from, say Christmas sales or economic activity tied to the school year of the travel season), the seasonal factors are estimates of average weather effects for each month, for example the average January decrease in new home construction n the Northeastern region of the US due to cold and storms. Seasonal adjustment does not account for abnormal weather conditions or for year-to-year changes in weather. It is important to note that seasonal factors are estimates based on present and past experience and that future data may show a different pattern of seasonal factors."
In a nutshell, this seems to say that the actual numbers do not fit nicely into mathematical models. They are unwieldy, and they do not take into account various factors that could make the results look worse (or better) than they really are[1]. As a result, in order to make the actual numbers fit the model and to try to account for these various factors, a "fudge factor" is applied.
It's tempting to point at this and scream "ZOMG THEY"RE FAKING THE DATA!!!!! The government is lying to us!!!!! Blood of Liberty! BLOOD OF LIBERTY!!! Information MUST BE FREE!!!! Free Assange! ASSANGE!!!!!!!!"[2]
But, before you work yourself into a "9/11 Truth" kind of conspiratorial hysteria, have a look at Report r539cy from the US Department of Labor. This report, run using the Labor Department's own data, is Unemployment Insurance Weekly Claims Data from 1/2/2010 through 12/11/2010[3]. Fifty weeks of data.

Initial Claims Continued Claims I.U.R Covered Employment
N.S.A S.F. S.A. N.S.A S.F. S.A. N.S.A S.A.
01/02/2010 645,446 141.4 456,000 6,013,891 123.7 4,862,000 4.6 3.7 130,128,328
01/09/2010 815,593 173.7 470,000 5,791,080 118.3 4,895,000 4.5 3.8 130,128,328
01/16/2010 652,327 133.3 489,000 5,602,357 115.8 4,838,000 4.3 3.7 130,128,328
01/23/2010 502,710 105.7 476,000 5,683,683 116.8 4,866,000 4.4 3.7 130,128,328
01/30/2010 533,320 108.8 490,000 5,683,530 118.5 4,796,000 4.4 3.7 130,128,328
02/06/2010 507,634 115.6 439,000 5,597,688 115.5 4,846,000 4.3 3.7 130,128,328
02/13/2010 478,235 99.9 479,000 5,546,408 115.7 4,794,000 4.3 3.7 130,128,328
02/20/2010 454,492 93.6 486,000 5,597,500 118.8 4,712,000 4.3 3.6 130,128,328
02/27/2010 471,256 101.1 466,000 5,538,966 118.0 4,694,000 4.3 3.6 130,128,328
03/06/2010 459,523 102.0 451,000 5,393,589 114.9 4,694,000 4.1 3.6 130,128,328
03/13/2010 434,424 95.6 454,000 5,344,610 114.5 4,668,000 4.1 3.6 130,128,328
03/20/2010 408,653 91.8 445,000 5,195,732 111.0 4,681,000 4.0 3.6 130,128,328
03/27/2010 408,204 92.3 442,000 5,040,758 110.5 4,562,000 3.9 3.5 130,128,328
04/03/2010 417,296 90.2 463,000 4,981,687 106.3 4,686,000 3.9 3.7 128,298,468
04/10/2010 510,927 106.4 480,000 4,924,111 105.6 4,663,000 3.8 3.6 128,298,468
04/17/2010 434,438 94.6 459,000 4,788,288 102.9 4,653,000 3.7 3.6 128,298,468
04/24/2010 426,412 94.5 451,000 4,668,932 100.9 4,627,000 3.6 3.6 128,298,468
05/01/2010 394,640 88.5 446,000 4,547,308 97.6 4,659,000 3.5 3.6 128,298,468
05/08/2010 409,759 91.8 446,000 4,469,365 96.0 4,656,000 3.5 3.6 128,298,468
05/15/2010 410,264 86.6 474,000 4,407,799 95.1 4,635,000 3.4 3.6 128,298,468
05/22/2010 406,875 87.9 463,000 4,377,445 92.8 4,717,000 3.4 3.7 128,298,468
05/29/2010 415,129 90.5 459,000 4,200,676 93.6 4,488,000 3.3 3.5 128,298,468
06/05/2010 395,396 86.2 459,000 4,308,561 93.8 4,593,000 3.4 3.6 128,298,468
06/12/2010 444,172 93.4 476,000 4,307,793 94.2 4,573,000 3.4 3.6 128,298,468
06/19/2010 423,438 92.2 459,000 4,330,835 93.4 4,637,000 3.4 3.6 128,298,468
06/26/2010 441,130 92.9 475,000 4,314,532 97.3 4,434,000 3.4 3.5 128,298,468
07/03/2010 468,456 102.2 458,000 4,394,779 93.3 4,710,000 3.5 3.7 126,763,245
07/10/2010 511,135 119.8 427,000 4,577,842 102.1 4,484,000 3.6 3.5 126,763,245
07/17/2010 502,473 107.3 468,000 4,568,163 100.0 4,568,000 3.6 3.6 126,763,245
07/24/2010 413,678 90.0 460,000 4,450,714 97.4 4,570,000 3.5 3.6 126,763,245
07/31/2010 402,135 83.4 482,000 4,333,520 96.5 4,491,000 3.4 3.5 126,763,245
08/07/2010 424,506 86.9 488,000 4,289,548 95.0 4,515,000 3.4 3.6 126,763,245
08/14/2010 404,503 80.3 504,000 4,219,684 94.2 4,479,000 3.3 3.5 126,763,245
08/21/2010 384,911 80.5 478,000 4,125,425 92.1 4,479,000 3.3 3.5 126,763,245
08/28/2010 383,111 80.2 478,000 4,138,969 90.5 4,573,000 3.3 3.6 126,763,245
09/04/2010 381,838 83.5 457,000 3,933,940 86.7 4,537,000 3.1 3.6 126,763,245
09/11/2010 341,664 75.5 453,000 3,930,917 86.7 4,534,000 3.1 3.6 126,763,245
09/18/2010 382,323 81.5 469,000 3,810,931 84.5 4,510,000 3.0 3.6 126,763,245
09/25/2010 372,536 81.7 456,000 3,779,896 83.8 4,511,000 3.0 3.6 126,763,245
10/02/2010 373,681 83.3 449,000 3,705,942 83.4 4,444,000 2.9 3.5 125,845,577
10/09/2010 462,667 97.4 475,000 3,715,958 83.0 4,477,000 3.0 3.6 125,845,577
10/16/2010 394,016 86.6 455,000 3,769,435 85.9 4,388,000 3.0 3.5 125,845,577
10/23/2010 408,488 93.4 437,000 3,759,357 85.7 4,387,000 3.0 3.5 125,845,577
10/30/2010 421,097 91.8 459,000 3,783,092 87.1 4,343,000 3.0 3.5 125,845,577
11/06/2010 452,657 103.5 437,000 3,735,928 86.4 4,324,000 3.0 3.4 125,845,577
11/13/2010 409,545 92.9 441,000 3,870,994 91.8 4,217,000 3.1 3.4 125,845,577
11/20/2010 464,813 113.4 410,000 3,665,773 85.7 4,277,000 2.9 3.4 125,845,577
11/27/2010 412,922 94.2 438,000 4,216,367 102.5 4,114,000 3.4 3.3 125,845,577
12/04/2010 585,678 138.4 423,000 4,062,518 97.5 4,167,000 3.2 3.3 125,845,577
12/11/2010 490,276 115.9 423,000 4,179,504 102.7 4,070,000 3.3 3.2 125,845,577
What's interesting in that 50 weeks of data, the Seasonally Adjusted figures are only better than the Non-Seasonally Adjusted figures 16 times[4]. The other 34 weeks, the SA figures were worse than the NSA figures.
So, before you whip yourself into a hysterical anti-authoritarian frenzy, ask yourself why - if there really is a conspiracy to "cook the numbers" - the numbers would be cooked so that almost 70% of the time the situation looks even worse than it actually is? I mean, really. That's almost the worst conspiracy ever.
Feel free to look at the NSA figures to see how many people are actually applying for UI in any given week. Then look at the SA figures to see how, assuming the correct seasonal factor was chosen[5], the given week compares to other weeks.
[1] I can't find the official list of these factors for first time jobless claims. I would assume that it includes things like the potential impact of seasonal positions ending in December (for the Christmas shopping season), massive one-time layoffs from a single large employer, companies that lay employees off for part of the year and then rehire them (**cough**Faygo**cough**cough**), and so forth.
[2] It's all right to admit that you had this response. I had it, mostly because I tend to assume that the government lies. All the time.
[3] I tried to run it through 12/31/2010, but apparently everything after 12/11 isn't in the database yet.
[4] 1/2, 1/9, 1/16, 1/23, 1/30, 2/6, 2/27, 3/6, 4/10, 7/3, 7/10, 7/17, 11/6, 11/20, 12/4, and 12/11.
[5] No guarantee, there. Economics isn't physics, after all.

No comments:

Post a Comment