Joe Weisenthal Dec. 4, 2009, 2:56 PM
Just about every time the monthly jobs numbers comes out, economic research firm TrimTabs comes out and slams the government’s methodology, usually honing in on the Birth/Death model of new businesses entering the market.
This week is no exception.
Frankly, we’re not sure what to make of their arguments. We’ve been hearing about this Birth-Death issue for a long time, but unless you believe they’re changing their methodology from month to month, then that issue only goes so far.
We welcome your thoughts.
TrimTabs’ Estimates 255,000 Jobs Lost in November, While BLS Reports a Decline of Only 11,000
BLS Revises September and October Results Down a Whopping 45%
Something’s Not Right in Kansas!
TrimTabs employment analysis, which uses real-time daily income tax deposits from all U.S. taxpayers to compute employment growth, estimated that the U.S. economy shed 255,000 jobs in November. This past month’s results were an improvement of only 10.2% from the 284,000 jobs lost in October.
Meanwhile, the Bureau of Labor Statistics (BLS) reported that the U.S. economy lost an astonishingly better than expected 11,000 jobs in November. In addition, the BLS revised their September and October results down a whopping 203,000 jobs, resulting in a 45% improvement over their preliminary results.
Something is not right in Kansas! Either the BLS results are wrong, our results are in error, or the truth lies somewhere in the middle.
We believe the BLS is grossly underestimating current job losses due to their flawed survey methodology. Those flaws include rigid seasonal adjustments, a mysterious birth/death adjustment, and the fact that only 40% to 60% of the BLS survey is complete by the time of the first release and subject to revision.
Seasonal adjustments are particularly problematic around the holiday season due to the large number of temporary holiday-related jobs added to payrolls in October and November which then disappear in January. In the past two months, the BLS seasonal adjustments subtracted 2.4 million jobs from the results. In January, when the seasonal adjustments are the largest of the year, the BLS will add anywhere from 2.0 to 2.3 million jobs. In our opinion, trying to glean monthly job losses numbering in the tens of thousands or even in the hundreds of thousands are lost in the enormous size of the seasonal adjustments.
In November, the BLS revised their September and October job losses down a surprising 44.5%, or 203,000 jobs. In the twelve months ending in October, the BLS revised their job loss estimates up or down by a staggering 679,000 jobs, or 13.0%. Until this past month, these revisions brought the BLS’ revised estimates to within a couple percent of TrimTabs’ original estimates.
The large divergence between the two results begs the question of what is causing the difference. While we don’t have an answer today, we will be poring over the data in an attempt to answer that question.
A comparison of TrimTabs’ employment results versus the BLS’ results from January 2008 through November 2009 is summarized below.
Several other employment related data statistics support the conclusion that the labor market is not as robust as the BLS is reporting:
Automatic Data Processing reported on Wednesday that 169,000 jobs were lost in November.
The Institute of Supply Management (ISM) Non-Manufacturing Survey reported that the majority of companies surveyed were still shedding employees.
The ISM Manufacturing Survey reported weaker employment conditions in November.
Weekly unemployment claims were 457,000 in the week ended November 27, 2009. While last week’s results were below the important psychological level 500,000, the weekly claims are still uncomfortably high and point to a contracting labor market.
The TrimTabs Online Jobs Index reported lower online job availability in the past three weeks.
The Monster Employment Index declined in November.
We will have the opportunity to truth our employment model estimates at the end of January 2010 when the BLS releases its annual benchmark revisions. The BLS revisions are based on actual payroll data for March 2009. The BLS revision is then divided by twelve to correct prior month’s data back to April 2008. We also use the March 2009 revisions to adjust our model inputs and make any necessary corrections.
For a complete analysis of the current employment situation and economic conditions, refer to TrimTabs Weekly Macro Analysis published this coming Tuesday, December 8, 2009