Biderman’s Daily Edge 2/3/2012: Is BLS Data Skewed?

Feb
03

The biggest headline for all financial media today is that the US economy added a much more than expected 243,000 jobs in January, and 446,000 jobs over the past two months. That is many more new jobs than our estimate of less than 50,000 for January and our estimate of 90,000 for December and January.

 

Our estimate of a slowly growing economy is based primarily upon daily income tax collections. Either there is something massively changed in the income tax collection world, or there is something very suspicious about today’s Bureau of Labor Statistics hugely positive number. We continue to check and recheck our analysis of income tax collections. We are aware that another service believes that incomes are growing faster than we do. So far we have not found any errors or discrepancies in our work, but if we do, we will let you know.

 

The BLS each month reports two data series, but only one jobs number is reported by the media. Actual jobs outstanding, not seasonally adjusted, are down 2.9 million over the past two months. It is only after seasonal adjustments – made at the sole discretion of the Bureau of Labor Statistics economists that 2.9 million less jobs gets translated into 446,000 new seasonally adjusted jobs for January and December.

 

No one I know has any idea as to how the BLS does this seasonal adjustment. BLS historic data is changed almost every month until the income tax returns for each year are available three years in arrears. In other words, the BLS currently has accurate data for 2008 and before.

 

I keep repeating that the BLS refuses to use the data embedded in income tax collections to be able to report real time jobs and wages. Why does it refuse? Could the reason it refuses to use real time data on jobs and incomes be because perhaps this jobs number is politically motivated? The entire world is looking at US job creation as a proxy on how well Obama is doing? Could the Obama administration be pressuring its economist employees to create the best possible new jobs number?

 

Obviously I am quite suspicious of the numbers that I see in today’s BLS press release.  Remember most financial journalists and even stock market strategists do nothing more than rewrite government press releases. So do not expect very few others to question the good news.

 

For those of you who care, look at Table B-1, Total Nonfarm Employment in today’s BLS press release attached to this video on our blog site. Start with the non seasonally adjusted table that shows that in November 2011, there were 133.172 million actual jobs. Actual jobs dropped by 220,000 jobs in December and actual jobs dropped an additional 2.7 million in January. Only as a result of unknown seasonal adjustments, could the BLS report 243,000 new hires in January.

 

Yes, the labor market contracts during the winter and expands in the spring and summer.  Could this number be manipulated? Of course it could. Is it? I don’t know.Am I the only suspicious soul out here? Hope not. Certainly feels lonely right now.

 

Charles Biderman
President & CEO
TrimTabs Investment Research

Tags:                                         

23 Responses to Biderman’s Daily Edge 2/3/2012: Is BLS Data Skewed?

  1. mary basel on February 3, 2012 at 8:38 pm

    great article…..

  2. jim on February 3, 2012 at 8:38 pm

    You are not alone. This is all BS or more accurately BLS.
    They can manipulate the numbers, but they cannot repeal the law of gravity.

  3. PF on February 3, 2012 at 10:02 pm

    Forget SA, what is the jobs picture YoY? And then as a percent of those between 18 and 65, or total labor pool. The other thing that BLS may be doing is folding those who are holding 3 jobs into three separate jobs, making it look like more people are working.

  4. David Gerson on February 3, 2012 at 10:29 pm

    If you look at the archived BLS report for Jan 2011 (for Dec to Jan), you can easily calculate that the seasonal adjustment factor they used was 2,934. If you use the same factor for Jan 2012 you get 245,000 which is less than 1% different from the 243,000 reported.

    No mystery on what the number they would use as a seasonal adjustment. How they determined they should use that number, I have no idea.

    Please explain if you are using seasonal adjustments in your tax receipt analysis.

    Thanks.

    David

  5. Rae on February 3, 2012 at 11:35 pm

    No, sir. You are not the only suspicious sole out here. Thanks for the explanation of the twisted numbers.

  6. James Dickey on February 4, 2012 at 4:29 am

    The BLS began using 2010 Census data for population figures as of Jan 2012. It was a note in the BLS press release.

  7. Steven Hales on February 4, 2012 at 8:20 am

    If you graph the not seasonally adjusted table b data for total non-farm employment you will see that seasonality returned to the data in 2010 and that within that seasonal variation there is a regular pattern related to holidays, christmas and summer vacation in the education sector. As well, there is a rising trend from 2010 to 2012. This Dec-Jan fall is no different than past dec-jan falls. My understanding is that by adjusting for these seasonalities we capture the rising trend (clearly seen in the raw data) and avoid false signals from seasonal volatility.

  8. Fabian on February 4, 2012 at 9:00 am

    “Am I the only suspicious soul out here?”

    No you are not. Seriously^^

    http://www.querschuesse.de/munchhausen-report/

    Please use google translater or maybe your stuff or contact the author^^

    Greetings from Germany

    Fabian A.

    The US still stucks in the jobless growth. The situation gets worther and worther. More and more people can not partizipate on the weak labour market.
    f.e.
    If you work 15 hours the week in Germany, than you will not be counted as unemployed! Nobody can seriously live with that.

  9. drdoc on February 4, 2012 at 10:35 am

    No, you’re obviously not “the only suspicious soul out here” – at least i did find the link to your blog on a german webpage:

    http://www.querschuesse.de/munchhausen-report/

    even if you don’t speak german anymore (“Biedermann” seems to be german),
    the chart-titles are in english, and the text is describing the charts.

    One question: why are there no charts about macro-economics within the mainstream-business-propaganda-media?

    exactly ….

    - strange times,
    - interesting times,
    -> head up, keep on doing – nice blog!

  10. L. Reichard White on February 4, 2012 at 2:53 pm

    Thanks CB!

    You don’t need to feel QUITE so lonely. All these figures are largely BS. Mr.Bernanke just admitted FED forecasts were really bad in his recent testimony.

    But without this column to remind me, they would have “had” me anyway. Thanks for keeping me on the straight and narrow!

    Health, happiness, & long life,
    L. Reichard White

    P.S. I think you may have meant, “So expect very few others to question the good news.” rather than “So do not expect very few others to question the good news.”

  11. Flatlaxity on February 4, 2012 at 4:34 pm

    Charles -

    Re your 2/3/2012 “Daily Edge”…

    While the assumptions behind the BLS’ seasonal adjustment factors may be in question, at least they seem to be consistently applied relative to last year’s for the months of Nov, Dec and Jan.

    Namely the effect, from Table B-1 from the BLS’ 2/3/2012 press release and that from 2/4/2011, of dividing the SA Establishment Nonfarm Payroll data by the NSA numbers, gives close results for year-year respective months.

    Every January the BLS does a big correction on its Birth/Death Model, the latest showing a job loss of 367,000 (NSA) versus just a negative 1,000 for the previous month. Normally this sizable January negative adjustment has an appreciable effect on the jobs gained/lost. But that’s not the case for last month.

    With the Secretary of Labor being under the President, and as you point out this being an election year, a potential for bias exists. John Williams of “Shadow Government Statistics” has mentioned examples of past Presidential involvement. And as you’ve stated, the Government’s data is constantly revised.

    I’m glad that you’re providing an independently formulated viewpoint with the use of daily tax data, regarding the all-important unemployment data otherwise only obtained from a sole Government source.

    Further, your video blogs are increasingly effective in making the public aware of TrimTabs’ independent analysis.
    - – - – - – - – - – - – -

    • Steven Hales on February 4, 2012 at 8:59 pm

      The yearly fluctuation in total employment is about 3 million peak to trough. That pattern has returned to the data. The seasonal adjustment is meant to flatten out these well known patterns. The upward trend has been apparent since January 2010.

  12. Charles Biderman on February 4, 2012 at 8:57 pm

    We do not use seasonal adjustment in our withholding analysis. We only use year over year. By the way, the US government is the only ones I am aware of that report month over month. Year over year eliminates the need for seasonal adjustments; which could be why the BLS and BEA resist using year over year reporting.

    • Steven Hales on February 4, 2012 at 11:26 pm

      Charles, Here are the yty values for total employment change for January (thousands)

      01/2003 -354
      01/2004 117
      01/2005 2004
      01/2006 2593
      01/2007 1990
      01/2008 888
      01/2009 -4285
      01/2010 -4246
      01/2011 1018
      01/2012 1936

      Here’s a selective analysis, the Bush Admin created 7 million jobs measured trough to trough.

      If the usual peaks to be reached in June and November continue to show rising employment yty will that be enough for you not to discount the BLS seasonal adjustments?

  13. Charles Biderman on February 4, 2012 at 9:00 pm

    We just heard from a good friend Jim Bianco who said:
    “The normal January weather adjustment is 424,000 people not at work due to
    the weather. But this January it was 206,000 less than in previous years
    and was apparently one of the lowest on record (this year’s Global Warming
    adjustment). “

    • Steven Hales on February 4, 2012 at 11:00 pm

      Charles,

      Here are the table B data for total employment changes dec-jan from 2003 to 2012

      01/2003 -2685
      01/2004 -2661
      01/2005 -2706
      01/2006 -2653
      01/2007 -2794
      01/2008 -3035
      01/2009 -3698
      01/2010 -2869
      01/2011 -2858
      01/2012 -2689

      Most of this is a result of seasonal hiring and firing in retail and wholesale and transport. There is nothing really nefarious going on here.

    • Larry on February 5, 2012 at 8:24 pm

      Wouldn’t a record low weather effect compared to years past lead to higher employment in your tax based series?

  14. Larry on February 5, 2012 at 8:01 pm

    Oh, come on. Seasonal adjustment is something any student of econometrics 100 can do. there is nothing strange going on here. The BLS method is more complex, but again any graduate student could do it. Furthermore, it is preferable to incorporate data as soon as possible rather than waiting up to a year to update the seasonal factors.

    If you doubt the numbers, you could argue that this year doesn’t follow the usual patterns and so the seasonal adjustment is too large. You could also claim that the seasonal patterns are gradually breaking down (ie christmas hiring plus firms reluctant to fire before Xmas followed by the end of seasonal jobs and pent up layoffs). Finally, BLS estimates (with little info) the number of people employed in newly created firms that are not on their Business Establishment List (that they use to sample from). Generally this can be a big number in the early stages of a recovery. This time round new start-up have been low, but maybe BLS over-estimates this.

    Your vague griping about BLS methods does you no credit. Do some research before your criticize.

  15. Charles Biderman on February 6, 2012 at 3:02 am

    The reason I am suspicious of the BLS number is due to wages and salaries after inflation declined 2.1%, adjusting withheld income and employment taxes collected by the US Treasury and reported daily. To come up with the -2.1% number we are estimating that bonus income is down 50%. Even if bonus income dropped by 100%, higher wages and salaries would indicate a job gain of no more than 140,000.

    One service says withholding grew over 8% in January. Yes, withholding grew by 8% only because there are two extra working days this January vs. Jan. 2011. Therefore that 8% gain is not due to more workers, rather more working days.

    We are suspicious of the number and given huge seasonal adjustments, if the BLS wants to fudge the number they can hide lots inside of a 2.7 million seasonal adjustment, plus a huge birth/death number and a warmer January resulting in much fewer jobs lost due to weather than last January.

  16. Will Morgan on February 8, 2012 at 12:07 am

    I work for a very large PC insurance Co. We just changed our pay policy. Our pay is now lagged 1 week resulting in our being paid for only 3 weeks in january 2012. Is this change being made at other companies? If so, it could be affecting your tax collection statistics.

  17. Will We Have a ‘Fair’ Election in 2012? on November 25, 2012 at 4:27 am

    [...] Obama and his MSM sycophants tout impressive, but misleading, economic figures.  (See here, here, here, here and [...]

  18. UGG オーストラリア on December 14, 2012 at 7:42 am

    last, UGG presents fashion. It’s attainable that as being a man you do not know many

Leave a Reply

Your email address will not be published. Required fields are marked *


Charles BidermanCharles Biderman is the Chairman of TrimTabs Investment Research and Portfolio Manager of the TrimTabs Float Shrink ETF (TTFS)

Biderman’s Practices of Success

I recently launched a new online course, Biderman's Practices of Success, on Curious.com (curious.com/charlesbiderman). The key to the practice of success is to be fully present in the moment and to be totally engaged in the important areas of life, particularly when we do not want to be. Read More.

Archives