Jobs raw and confusing

Jan	143526	146552		3026	608	566
Feb	144134	147118		2984	956	517
Mar	145090	147635		2545	677	952
Apr	145767	148587		2820	631	762
May	146398	149349		2951	706	296
June	147104	149645		2541	161	77
July	147265	149722		2457	-618	-494	Teachers laid off
Aug	146647	149228		2581	294	-248
Sept	146941	148980		2039	995	736	Teachers rehired
Oct	147936	149716		1780	-270	50	Construction slowdown
Nov	147666	149766		2100	-476	-63
Dec	147190	149703		2513	-638	-666	Christmas hires laid off

I went through jobs not seasonally adjusted, Table A13 in monthly BLS employment survey.  So the numbers above are simple estimations based on the current survey. 

The two left columns are actual jobs for year 2014 and 2015, so I am comparing two years. Not much, bur two out of the five that make the expansion.  The two right columns are the non-seasonally adjusted job gains, for the following month.  I added a column of probable causes of the seasonality.

First, we lose about 600,000 jobs each January, so the adjuster has turned a negative 666 thousand into a positive 150 thousand. Also note that the warm winter in 2015 mean fewer construction lay offs.  And the number of teachers laid off for the summer seems normal.

So, how did we get 150k new jobs instead of 190k, when starting with a negative 666k?  Well, we make this bold assumption the adjuster is a zero mean filter, and maintains the same response, until otherwise notified.  But just looking, this jobs report is ambiguous, it really tells us nothing until around April.