What's Inside the BLS Magic Black Box?

Inquiring minds have asked "I'm curious if you can translate the -339 birth/death adjustment this quarter. It seems completely out of context."

I believe I can shed some light on the overall process and the finalized numbers. However, the black box itself is impenetrable.

Here is the process: The BLS takes seasonally unadjusted numbers for jobs and seasonally unadjusted model-derived birth-death numbers and meshes them inside a magic black box. The box slices and dices the numbers, then spits out the seasonally adjusted "establishment number".

As a side note, when tossing the unadjusted numbers inside its black box, rumor has it the BLS employees collectively sing Oh Ho Ho It's Magic. I cannot confirm that rumor officially. Nonetheless, here is a fitting tribute.



In the monthly "Ta Da!" the BLS cleverly gives us the seasonally adjusted commingled result (that's the official monthly establishment number). The BLS also gives us the not seasonally adjusted birth death model number.

It makes as much sense to subtract those numbers as it does to subtract 3 bananas from two oranges, yet people keep trying.

It would help if the BLS divulged the black box methodology, but it doesn't. This rightfully arouses suspicion.

Interestingly, the BLS cannot give us the seasonally adjusted birth death number because it does not have the number to give! It's all crunched together in the impenetrable black box.

January Effect

The January phenomenon is interesting. The birth-death number for January is always a huge negative number. Do businesses have a propensity to go out of business precisely in January (and only January), come hell or high water, recession or not?

It must be so because for 5 years (until November 2010), the only other month I recall seeing negative a birth-death number is in July.

Flashback December 4, 2009: Jobs Contract 23rd Straight Month; Unemployment Rate Drop to 10.0%

Birth Death Model Revisions 2008 - April thru December



click on chart for sharper image

Birth Death Model Revisions 2009



click on chart for sharper image

In 2008 and 2009, smack in the middle of the recession, the BLS said more jobs were created by new businesses than lost by businesses going out of business, every month but January.

Here is a table I produced that shows the same thing in 2007 .



Presumably, NET new business job expansion occurs like clockwork (except for January). This mysteriously changed for the first time I recall in years of watching, in November of 2010.

Flashback December 3, 2010: After All the Hype, Jobs Up an Anemic 39,000; Unemployment Rises to 9.8%; Quicksand of Stimulus

Birth Death Model Revisions 2010



click on chart for sharper image

I am actually shocked to see birth-death adjustments not only back in the solar system, but also back on planet earth. I cannot recall the last negative number in any month but January or July.

Curiously, smack in the middle of an expansion, we now have one isolated case of net business contraction, although we had NET new business job expansion every month the entire recession, except January.

Hold Your Horses

Please note the BLS revised their model this month, and something entirely new and exciting is happening: The latest revisions show we now have net birth-death job destruction in January, July, September, and November.

Let's take a look.

Flash Forward February 4, 2011: Labor Force and Unemployment Statistical BS

Birth Death Model Revisions 2010 (as reported last month)



Birth Death Model Revisions 2010 (as reported this month)

Once again, this is all massaged inside the BLS black box. All we know is the whole damn thing did not make any sense then, and these revised negative numbers out of the blue now sure do not correlate with anything we have seen before.

I suppose all those self-employed business owners selling trinkets on E-Bay may have collectively tossed in the towel this year, but otherwise it is hard to explain.

Quarterly Benchmarks

The BLS has taken it's black box to the next magic level - quarterly instead of annual benchmarks. Please consider Introduction of Quarterly Birth/Death Model Updates in the Establishment Survey

Effective with the release of preliminary January 2011 employment estimates in February 2011, BLS began updating the Current Employment Statistics (CES) net birth/death model component of the estimation process more frequently, generating birth/death residuals on a quarterly basis instead of annually. This will allow CES to incorporate Quarterly Census of Employment and Wages (QCEW) data into the birth/death model as soon as it becomes available. This more frequent updating will help to reduce what is known as the "post-benchmark revision" in the CES series. There will be no change to the timing or frequency of the current CES monthly or annual releases. Benchmarking to the QCEW will continue to be done on an annual basis. The first monthly estimates to be affected by this change will be for the reference month of April 2011.

2003-2010 Birth-Death Model Annual vs. Quarterly Benchmarks



In fairness to the BLS, the latest revisions (assuming they are correct), only subtracted 405,000 jobs in 2009. That is a decent-sized revision, but not a massive one. This lends credence to the idea that too much attention is given to the birth-death numbers.

Regardless, the quarterly process does seem much more reasonable than anything we have seen before.

Moreover, it would have been interesting to see month-by-month revisions for 2009. Perhaps we would have seen some more negative months. That may have made some of the 2009 months look as if they were from this solar system instead of Bizarro World.

Summation

The BLS refuses to describe the magic in its black box magic. Suspicion will rightfully remain until that happens.

The way the BLS reports these numbers (purposely or not) makes it hard to do math on the monthly numbers. This leads to confusion and wild assumptions.

Quarterly benchmarks appear to be a significant improvement.

Looking back at the data presented above, year in and year out (even the positive years), the weakest and therefore the most likely negative birth-death months going forward are in order January (a lock), followed by November, September, and July.

The January effect may be explainable, the rest is somewhat curious.

Mike "Mish" Shedlock
http://globaleconomicanalysis.blogspot.com
Click Here To Scroll Thru My Recent Post List