Fun with the BLS Birth/Death Model
It's been a while since I looked at the BLS Birth/Death Model.
I took a look just for grins and I challenge at least one BLS model assumption.
The birth/death model is for the birth and death of businesses, not employees. The BLS assumes there are business startups each month that they do not know about. There are also corporations that go out of businesses and everyone loses their job.
The BLS estimates the number of newly employed and newly unemployed people due specifically to the birth and death of businesses, but only on a seasonally unadjusted basis.
Let's take a look, with a spotlight on construction employment.
2016 Net Birth/Death Adjustment
Numbers Not Comparable
Those numbers are not comparable. One set of numbers is seasonally adjusted but the other is not.
Yet, it's rather curious that 74,000 construction jobs were allegedly created in the last two months at the same time construction lost employment for two consecutive months for the first time since 2012.
Evey month in my report on jobs, I write ...
Birth Death Model
Starting January 2014, I dropped the Birth/Death Model charts from this report. For those who follow the numbers, I retain this caution: Do not subtract the reported Birth-Death number from the reported headline number. That approach is statistically invalid. Should anything interesting arise in the Birth/Death numbers, I will add the charts back.
Seasonal Adjustment Mess
The problem in untangling this mess is the way the BLS does its calculations.
The BLS applies the non-adjusted birth death numbers to the total non-adjusted numbers, then seasonally adjusts the total.
I cannot tell you how much the birth/death model affected the seasonally-adjusted overall total, nor can anyone else.
Not even the BLS can tell you.
The overall effect of the birth/death numbers on the overall numbers is not as big as it seems. That we can show.
In March, the BLS estimated there were 6,560,000 construction employees. The BLS estimated another 40,000 due to the birth of new businesses.
Rounding to three decimals, the birth/death adjustment is about 0.610 percentage points of the total number of construction jobs.
- Not Seasonally Adjusted, the BLS estimates 6,560,000 construction employees.
- Not Seasonally Adjusted, the BLS estimated another 40,000 construction employees due to the birth of new businesses.
- After seasonally adjusting everything, the BLS says construction employment went down by 15,000.
- Seasonally adjusted, the BLS estimates 6,645,000 construction employees.
- Construction employment would have gone down more had the birth/death model not added jobs. By how much, we do not know.
Overall, the BLS says the birth/death model created 224,000 jobs. The BLS estimates nonfarm employment at 144,592,000.
Rounding to three decimals, the percentage of jobs the model added to the overall total is a mere 0.155%.
I do not know how either of those percentages translates to the overall seasonally adjusted numbers (and no one else does either).
Regardless, I question whether any net construction jobs were genuinely added by the birth of businesses at a time when construction spending was down for the month and construction employment was down two months straight.
If the birth/death model added 0.6% to construction employment when instead it should have subtracted, there may be a meaningful error.
- Retail Hiring Slowdown Coming Up: Construction Spending Plunges 1.8%
- Construction Employment Declines Back-to-Back First Time Since May 2012; Questions of the Day
- Fed Hiking Not: Payroll Jobs +38K, Employed +26K, Labor Force -458K, Revisions -59K