An Invitation to Visitors to the Gillespie Research Website
If you are new to this site, please look around while you are here. If you like what you see, be sure to sign up for a free trial subscription. You will find more about this at Information about Free Trial Subscriptions.
______
Summary
In the August edition of Shadow Government Statistics, John Williams calculated the July gain in payroll employment to be 44,000, versus the 207,000 reported last week by the Bureau of Labor Statistics. In this missive, John explains how he came up with this significant discrepancy.
Introduction
Gillespie Research Associates finalized its joint-venture research relationship with John Williams late last year with the strong conviction there were many people -- an expanding number, to be sure -- who knew there was a good deal about reported government economic data that simply was not sensible. John's knowledge of this area, and his ability to analyze and articulate what is going on, are right up there with the best in the field. Thus, the enthusiastic response to his work, dispensed in his monthly newsletter, "John Williams Shadow Government Statistics," has not been at all surprising.
The newsletter's August edition was published earlier this week, and it contained the following observations about the payroll employment data released by the Bureau of Labor Statistics on August 5th:
"July's relatively strong reported payroll numbers reflected unusually distorted seasonal factors, but met the political needs of the administration. In theory, seasonal factors distorted in one direction eventually have an offset within a given year's adjustments. Such no longer is the case with the jobs data, since the Bureau of Labor Statistics has been calculating and revising monthly seasonal adjustments, each month, as it goes along. July was the second consecutive month of upside seasonal distortion, the third such result in the last four months.
"...Moving in tandem with household employment, the June payroll survey showed a seasonally-adjusted gain of 207,000 ... despite a negative bias factor that subtracted 76,000 jobs from the underlying unadjusted data (80,000 jobs were subtracted last July).
"Once again, patterns of year-to-year growth in the payroll series have been highly irregular and highlight unconscionable games being played with the month-to-month seasonal factors...Reported on an adjusted basis that is consistent with the unadjusted annual growth, July payrolls would have risen by 44,000, some 163,000 jobs less than the official estimate."
_____
John Williams Elaborates on How He Arrived at the 44,000 Figure
In hindsight, I believe I probably should have gone further in the August newsletter ["John Williams' Shadow Government Statistics"] in explaining how I arrived at a 44,000 July jobs gain, net of seasonal distortions. Of course, this result was far short of the 207,000 gain in payroll employment reported last week by the US Labor Department. Therefore, I will expand on the reporting problem and how I calculated my estimate.
Unadjusted payrolls in July usually drop off sharply from June, so some seasonal adjustment indeed is needed to balance the picture. Seasonal adjustments tend to be fairly consistent year to year, and there is no unusual factor I know of that has distorted either June or July of 2005 versus June or July of 2004, other than the BLS arbitrarily calculating and revising its seasonal factors this year, month-to-month, instead of
calculating them in advance, as they did last year.
Looking at the numbers below, however, it is clear the 2005 seasonals have been altered from 2004, in favor of boosting 2005. In June to July 2004, unadjusted payrolls fell by 1.143 million (including the 80,000 subtracted in July with the birth/death model fudge factor), but adjusted jobs rose by
83,000.
In the same period of 2005, unadjusted jobs fell by 1.201 million (including a negative 76,000 fudge factor), but the adjusted jobs rose by 207,000.
Unadjusted, 58,000 more jobs were lost in 2005 than in 2004, but after seasonals, 124,000 more jobs were "created." That is a positive swing of 182,000 jobs, thanks largely to changes in the seasonals, but there also was a minimal contribution from the overall growth in the population and universe of employed individuals.
If the data were properly adjusted for seasonal factors, year-to-year growth would be the same on both an unadjusted and adjusted basis, and I find that approach to be the simplest in estimating the effects of gimmicked seasonals.
As noted in today's SGS text, the adjusted series showed an uptick in annual growth, while the unadjusted showed a minor downturn. In theory, the cleanest numbers are the unadjusted numbers, and applying the unadjusted annual growth rates to the adjusted 2004 numbers generates a new set of 2005 numbers that allows the best estimate I can come up with as to the impact of
the inconsistent seasonal factors.
Applying the unadjusted annual growth rates of 1.66% and 1.63% for June and July 2005 respectively to the seasonally adjusted payroll levels of 131.479 million and 131.562 million for June and July 2004, recasts the June and July 2005 numbers on a largely consistent basis with last year's seasonal factors.
The recast June 2005 is 133,662, with July 2005 at 133,706, an increase of 44,000. That is the number I used in the newsletter, a number 163,000 shy of the official report.
---------------------------------------------
REPORTED PAYROLLS (In Thousands)
---------------------------------------------
Month/ Month/
Year Result Year Result Change
---------------------------------------------
Seasonally Adjusted
-------------------
06/04 131,479 07/04 131,562 +83
06/05 133,579 07/05 133,786 +207
------- -------
Change 2,100 2,224
1.60% 1.69%
---------------------------------------------
Not Seasonally Adjusted
-----------------------
06/04 132,527 07/04 131,384
06/05 134,732 07/05 133,531
------- -------
Change 2,205 2,147
1.66% 1.63%
---------------------------------------------
Using Unadjusted Year/Year Growth
Patterns Against 2004 Adjusted Data
-----------------------------------
Recast Recast
06/05 133,662 07/05 133,706 +44
---------------------------------------------
The 44,000 is an estimate that shows the impact of an unjustifiable shift in seasonal factors, enough of a shift to set off alarms of possible manipulation.
Given the combination of factors mentioned in my prior newsletter, predicting an outright monthly payroll decline for July was the best bet I could give you. What happened with the actual reporting appears to have been more manipulative than fundamental. Current politics can take on a heavy roll in these numbers, and open manipulation may be back on the table
for the first time in quite a while.
___
NOTE: If you are not familiar with John Williams' Shadow Government Statistics and would like to know more about it, leave your name and e-mail address behind at the "Contact Us" section of that publication's website. We will be happy to provide you with some background information.
_____
|
