Flash Update
A L E R T
October 9, 2006
__________
Political Manipulation of Labor Data Kicks into High Gear
September Payrolls Fell 40,000 Using Consistent Seasonal Adjustments
Help-Wanted Advertising Plunges Again
__________
Please note that the SGS Ongoing M3 and Alternate GDP and Dollar measures have been updated on the Alternate Data tab of the home page www.shadowstats.com.
The broad outlook for a deepening inflationary recession remains in place. The September employment report showed severe deterioration, despite a number of reporting gimmicks. Beyond any reporting shenanigans, economic growth has been moving rapidly to the downside in economic releases of the last week or so. Faced with an electorate that is in economic pain, the Bush Administration has tried to make the bad numbers disappear for a while, but the results have been mixed, so far. This is despite the comparative annual boosts starting to show up in data from the effects of last year's terrible hurricane season. The heavily touted annual gain in September retail stores sales is a prime example of such an effect.
Moving into the realm of the unbelievable, the Administration claims that it inadvertently understated jobs growth in the year ended March 2006 by 810,000. This may give George Bush bragging rights as to having created more jobs, but, like his father claiming an early end to the 1990 recession, the President is at risk of appearing to voters as being out of touch with economic reality.
Employment/Unemployment. Seasonally-adjusted September payrolls rose in the month by 51,000 +/- 106,000, after August's 188,000 gain (previously 128,000). Aside from being statistically indistinguishable from a monthly contraction, the September reading was the sixth that averaged below 120,000. That said, the September payroll numbers had several strange features.
Unusual, but of as yet undetermined significance, the unadjusted August payroll level was revised upward by an unusually large 99,000, but only 62,000 of that showed up in the seasonally adjusted August number. The Bureau of Labor Statistics (BLS) readjusts the seasonal factors each month, which allows it to report most anything it wants to report, but here is where the September data simply go awry.
Keep in mind that a year ago September, the Gulf Coast was devastated by Hurricane Katrina. While the BLS now shows monthly payroll growth slowing from 175,000 in August 2005 to 48,000 in September 2005, the BLS never reported the bulk of lost jobs. Even so, looking at August 2006 and September 2006 against that year ago period, one would expect that a growing economy would spike the relative annual jobs growth for September. With September 2006 being compared to an artificially-depressed period, however, growth did not increase; it softened instead.
On a not-seasonally-adjusted basis, August 2006 payrolls were up 1.34% against last year, while September 2006 was up by only 1.27%! This means that without a devastating hurricane, September took a relatively bigger hit in 2006 than it did in 2005. September 2006 annual growth was consistent with a hit from something worse than Katrina, but there was no hit other than a deepening recession. Since these jobs numbers are, in theory, unadjusted, one cannot blame the shockingly-weak results on bad seasonal factors.
When data are seasonally adjusted properly, year-to-year growth in the adjusted and unadjusted monthly data should be virtually identical. In September, they were not. The unadjusted annual growth rates detailed above, when applied to the seasonally-adjusted numbers, indicate that monthly payrolls should not have risen by 51,000 in September, but rather they should have fallen by 40,000! So how did jobs rise by 51,000?
Somehow the seasonally-adjusted annual growth rates ended up as 1.323% and 1.325%, respectively, for August and September 2006. Even with adjusted annual growth holding at the same level in September as August -- instead of September spiking against Katrina's impact -- the data suggest something close to collapsing economic activity.
The problem here likely is that the monthly data have become about as worthless as the quarterly GDP series. My speculation is that someone at the BLS realized too late that they were going to have consistency problems with the payroll data related to Katrina and last year's other hurricanes. Last minute revisions and unusual seasonals were used to generate positive jobs growth for September, while keeping the annual growth rate from falling. What was seen was not normal or regular reporting. One does not have consistency problems with honest data. Whatever was gimmicked was not done well.
Adding to the extraordinary nature of the September jobs report, the BLS announced that its annual benchmark revision for March 2006, to be revised and reported with the January 2007 jobs data, would be an addition of roughly 810,000 payroll jobs. Although historical revisions have been worse, the BLS announced that it "currently is researching possible sources for this larger-than-normal expected benchmark revision."
The benchmarking process has been seriously flawed for years, and this change is not credible. Consider that the involved period encompasses the missing of several hundred thousand jobs lost along the Gulf Coast. If a cause for the "undercount" cannot be found, the solution will be to add roughly 67,000 to 100,000 extra jobs a month to the birth-death bias factors, so as to overstate jobs growth even further, on a regular basis. Remember, the only political sin for a government statistical agency is to underestimate economic growth or to overestimate inflation.
On a happier, voter-friendly front, the household survey showed the seasonally-adjusted unemployment rate easing to 4.58% +/- 0.2% in September, from 4.69% in August, a statistically insignificant decline, but nonetheless in time for the election. U-3, on an unadjusted basis, eased to 4.4% in September from 4.6% in August. The broader U-6 rate was an adjusted 8.0% in September, down from 8.4% in August. Adjusting for the impact of the elimination of the counting of millions of discouraged workers, defined away by the Clinton Administration, the broadest, traditional unemployment rate remains about 12%.
Help Wanted Advertising Index. Confirming the extremely bleak employment picture, the August help wanted advertising index dropped to 31, from 32 in July, and from 34 in June. The August level is the lowest reading since April 1961. If the index drops another point, comparisons will shift from the Kennedy Administration to the Eisenhower Administration. While the series never recovered from the 2000 recession, its recent plunge -- annual change is down 15.7% on a three-month moving average basis -- is signaling a renewed contraction in economic activity.
GDP. The "final" estimate revision of seasonally-adjusted annualized quarterly real (inflation-adjusted) GDP to 2.56% from 2.90% had little obvious relationship to underlying economic activity. The changes primarily were in inventories fudge factor. Parallel GNP growth slowed to 2.34% from 2.60%. This series remains more a tool of political propaganda than an indicator of economic activity.
Inflation. In Mr. Bernanke's fantasy world, where the core rate of inflation used in deflating the appropriate portion of personal consumption expenditure (PCE) is the ideal measure of price changes, the news was not good. While the aggregated PCE annual deflation rate eased to 3.2% in August from 3.4% in July, the core rate increased to 2.5% from 2.3%. It seems that energy prices are beginning to permeate the broad spectrum of economic activity.
Other Numbers. August new orders for durable goods were down 0.5% for the month, with annual growth slowing to 3.6% from 10.3% in July. Annual money supply growth in September was weaker for M1 and M2, but the SGS estimate of M3 annual growth showed a minimal increase to roughly 9.2%, from 9.1% in August.
CAUTION. As discussed in the last newsletter, September economic and inflation data will be skewed both by year-ago Katrina effects and likely pre-election political manipulation. Economic data will tend to spike, and inflation will fall, sharply. The process will reverse in post-election reporting. Further details on all of the preceding will follow in the regular October SGS.
October's "Shadow Government Statistics" is scheduled for release on Wednesday, October 25, 2006. The posting of the next SGS, as well as any Interim Updates or Alerts, will be advised immediately by e-mail.