BLS Fudging Numbers

For a few months now I’ve been ranting about how there is no way that the Unemployment numbers are right because they don’t match up with the employment numbers. A while back, I got an angry email from a subscriber saying that he “worked for the Treasury Department for 35  years and there was no way the BLS fudged any numbers.”  So I responded and said I’d be happy to post his rebuttal if he could explain how those numbers could possibly match up… I’m still waiting for his response.

In February, I got an answer, not directly mind you, but when I looked at the January Employment numbers they have magically changed all the way back to July of 1991. Yes, “unadjusted” employment numbers have been adjusted, changed, fudged, manipulated whatever you want to call it.

In July 1991 and preceding, the numbers are the same. In August 1991, the difference is minuscule only 1000 mysterious jobs have been added. But the difference continues to grow. And by December of 2012 the old number was 134.822 million and the new number is 135.545 million so somehow magically the BLS has created 723,000 jobs out of thin air.

Who do they think they are to create  Jobs out of thin air, the FED?

We know the FED can create money out of thin air, so maybe the BLS can do the same with jobs?

By the way the BLS does say that the numbers are “Preliminary” for 2 months so theoretically they could legitimately adjust the November and December numbers… but going all the way back to 1991 is a bit much! But, if I were to fudge numbers that is exactly what I would do …gradually adjust them from way back so there isn’t a sharp disconnect between the old series and the new.

Here is a chart of the old employment vs. new fudged numbers since 2000.

Note where the 738,000 jobs difference occurs.

 Click for larger image

I contacted the Bureau of Labor Statistics and asked why the numbers were changed. I received a curt reply pointing me to the “answer” on their website. It starts out with all this mumbo-jumbo about this minor change from April 2010 – March 2011 and goes on for a couple of paragraphs about it and then they slide in one key sentence with no fanfare “oh by the way we changed all the employment numbers back to 1991 too.”

Here is the BLS Response:

Reconstructed data

In order to provide a continuous and comparable employment time series, CES decided to reconstruct the history for one CES series, 55-524126 Direct property and casualty insurers, impacted by the changes to noncovered employment incorporated with the 2011 benchmark (see www.bls.gov/ces/cesbmart11.pdf). CES determined that the history of this series did not capture a significant amount of noncovered employment and instead showed a level shift in the series from April 2010 to March 2011, where the noncovered employment, calculated from County Business Patterns (CBP) data, was first included. CES uses CBP data to calculate the majority of noncovered employment. The CBP — which draws from Social Security filings and other records which do include those employees not covered by UI tax laws — is lagged in its publication by approximately two years (e.g. in 2012 the 2010 CBP data was published). For more information about calculating noncovered employment, see bls.gov/web/empsit/cestn.htm#NCE

CBP data was used to reconstruct AE data for CES series 55-524126 back to 1991 in order to incorporate the noncovered employment in the history. From the reconstructed AE data, the remaining published data types were produced using the previously published sample ratios.

This reconstruction also resulted in revisions back to 1991 for the associated aggregate series, including but not limited to Total nonfarm, in the associated data types.

So even though according to the BLS they survey employers to find out how many people are employed they somehow figure they aren’t counting everyone, so they need to fudge the numbers except they call it “reconstructing” because they somehow aren’t counting all the “Property and Casualty insurers”. So back in 1991 this was 1000 and now they are missing 3/4 of a million of them.  Sounds like the Property and Casualty business must be booming to go from 1,000 to 738,000 since 1991. (Emphasis mine- the BLS did not bold the statement.)

 The following table includes all the original numbers and the new changed employment numbers from July 1991 through the current data. Note that in July 1991 and prior the numbers are identical. This data series is called “Not Seasonally Adjusted Numbers” Series CEU0000000001 but it seems a bit hypocritical to call them unadjusted at this point.

So there you have it, the BLS decided to “reconstruct” the employment numbers.  Gives you lots of confidence in their reporting doesn’t it? Plus this new fudge factor is now “baked in the cake” so all new numbers after January will have it already fudged in so we will never know what the real numbers are.

Webmasters: Feel free to repost this article, all we ask is that you link back to this original article on UnemploymentData.com.

Image courtesy of Salvatore Vuono / FreeDigitalPhotos.net