This morning I posted Jobs Rise Less than Expected 160K; Employed -316K, Fulltime Employment -196K.
The number in question is Fulltime Employment at -196,000. Zerohedge arrived at -253,000.
Attempting to sort things out between the table I used and the one ZeroHedge used, I noted the numbers don’t add up.
There is a discrepancy in part-time employment numbers this month of 1,366,000 between two different tables.
With that introduction, let’s dive into Household Table A as compared to Household Table A9.
Table A9
Employment dropped by 316,000. That’s the number I used in my report. It matches the number in Table A.
However, please note that the sum of the parts does not equal the whole.
- Full-time + part-time = 150,991
- Total employment = 151,004
Numbers get even further convoluted if one adds up the change in full-time workers plus the change in part-time workers.
- Change in full-time + change in part-time = -274,000
- Change in employment = -316,000
Table A
The -316,000 decline in employment between table A and Table A9 matches. Everything else is wildly off.
Part-Time Math
Mathematically speaking, you might think that one would be able to add part-time employment for economic reasons to part-time employment for non-economic reasons and arrive at something at least close to total part-time employment, but one would be very wrong.
For April, the sum of part-time for economic reasons plus part-time for non-economic reasons totals 26,431.
From Table A9, we see part-time employment is 27,719. Math shows the sum of the parts is less than the whole by 1,366. Since the numbers are in thousands, we are talking about a 1,366,000 employees (about 5% of the part-time total).
I arrived at a decline in full-time employment of 196,000 by starting with -316,000 and subtracting the part-time subtotals.
A call to the BLS straightened things up.
Because of the way the BLS does seasonal adjustments, you cannot expect anything to add up. And it won’t, perhaps by millions.
For calculating changes in full-time employment, one cannot make logical assumptions from Table A, which otherwise has most of the numbers I post on monthly.
The BLS’s best guess (not that the guess is any good) is that full-time employment declined by 253,000 this month, not 196,000.
As for the number of part-time employees and breakdowns thereof, the numbers are so wildly off it’s safe to assume +-2 million either way, more or less … I think.
Mike “Mish” Shedlock
.
.
…. Summertime… Summertime… Summertime
.
.I can see the grandkids,
and don’t need to be an Uber driver to pass my time !
.
.
. See you next fall.
Why not use unseasonal numbers and compare it year on year instead? This would mitigate the need for seasonal.
You cannot compare month-to-month except seasonally!
Mish
Seasonal correction is a license to cheat. Merely superimpose the annual graphs Comparison becomes obvious and easy.
NOTE: Detail for the seasonally adjusted data shown in this table will not necessarily add to totals because of the independent seasonal adjustment of the various series. Updated population controls are introduced annually with the release of January data.
BLS
Barack’s
Labor
Statistics
Where nothing makes any sense.
Pingback: Weekend Open Thread – Blogs For Victory
The BLS invented Common Core. ;o)