Even BLS Admits Birth-Death Model Skews Unemployment Numbers During Recessions → Washingtons Blog
Even BLS Admits Birth-Death Model Skews Unemployment Numbers During Recessions - Washingtons Blog

Monday, October 5, 2009

Even BLS Admits Birth-Death Model Skews Unemployment Numbers During Recessions

Barry Ritholtz and other writers have hammered the Bureau of Labor Statistics' (BLS) faulty birth-death model for years.

But as Ritholtz notes today, that ignorance about the inaccuracy of the birth-death model during recessions might finally might be lifting:

It is finally being recognized in the mainstream as the massive data distorter that it is. The latest BLS analysis and data revision shows that during 2008, the Birth Death adjustment caused NFP payrolls to be significantly under reported.

NYT’s Floyd Norris:

“It now appears that during the first half of 2008, when the recession was getting under way, job losses averaged 146,000 per month. That is nearly three times the average of 49,000 jobs shown in the initial estimates.

How did the government get it so wrong?

The official job numbers are based on a monthly survey of employers, augmented by something called the “birth-death model,” which factors in jobs assumed to have been created by employers who are too new to have been included in the survey, and subtracts jobs from employers assumed to have failed and therefore not responded to the latest survey.” (emphasis added)

Triple the job losses than reported, and right at a crucial part of the economic cycle! Is it any wonder policy response from central bankers and pols was so off? At the most crucial time, they failed to see the oncoming headlights, because they were lost in a fog of data so massaged as to have it completely and totally misrepresent reality.

About time this nonsense was recognized for the bullshit it is. We need to have BLS needs to toss out the 2003 modification to the B/D. We should get back to actually counting, rather than imagining, jobs.

Indeed, even BLS itself is now admitting the birth-death model has problems. As Bloomberg notes:

From April 2008 through December, the tax records showed the Labor Department’s figures overestimated payrolls by about 150,000, said Chris Manning, the national benchmark branch chief at the Bureau of Labor Statistics. That implies the estimates missed the mark by about 675,000 in the first quarter of this year...

“In this period of steep job losses, the birth/death model didn’t work as well as it usually does,” Manning said in an interview. “To the extent that there was an overstatement in the birth/death model, that is likely to still be there.”

The model added about 184,000 jobs to the payroll total last quarter compared with a 135,000 increase in the same period in 2008, before the financial crisis deepened with the collapse of Lehman Brothers Inc.


  1. All the government numbers have been buggered go to Shadow Government Statistics and get the numbers as they were reported in yesteryear.

  2. It's a mess out there:
    here is a map of Florida Unemployment in August 2009 (BLS data)

    versus Florida Unemployment Levels 1 year ago


→ Thank you for contributing to the conversation by commenting. We try to read all of the comments (but don't always have the time).

→ If you write a long comment, please use paragraph breaks. Otherwise, no one will read it. Many people still won't read it, so shorter is usually better (but it's your choice).

→ The following types of comments will be deleted if we happen to see them:

-- Comments that criticize any class of people as a whole, especially when based on an attribute they don't have control over

-- Comments that explicitly call for violence

→ Because we do not read all of the comments, I am not responsible for any unlawful or distasteful comments.