jobs data is quite confusing. what is “seasonal adjustment” anyway?
Seasonal adjustment usually attempts to correct for “expected” changes in employment level during the year. For example during the Christmas holiday season, lots of part-time workers are usually hired and then let go after New Year’s. To get a better sense of how the underlying economic trends go, BLS attempts to figure out how to strip out these sorts of yearly events and get to the “real” number of net jobs created or lost. The problem is that they have to use some cagey assumptions about how strong these yearly fluctuations are in a given year and can be slow to reflect changes in employment trends.
“Seasonal adjustment is a statistical technique that attempts to measure and remove the influences of predictable seasonal patterns to reveal how employment and unemployment change from month to month”
http://www.bls.gov/cps/seasfaq.htm
Remember high school chem lab? Same deal. The technical term is "fudge factor".
More people work during the summer (farms, teens at fast food, construction - not much here) and around Christmas (retail) so the BLS compares this September vs. previous Septembers. The loss of government jobs has to be K-12 teachers, mostly not getting hired and early retirement.
Teachers rarely quit, as a K-12 teacher in suburban Columbus, OH starts at about $40K and moves up to more than $75K. It’s expensive for a teacher to move, as they lose seniority, so many/most teachers are at one district their entire careers.
Since teacher salaries rarely move down (only in extreme local conditions) and real estate taxes are designed to collect an amount (meaning if all houses drop in appraised value by 50%, the taxes remain the same), a district only loses money from the state, or when properties go vacant, so that no one is paying real estate taxes. Figure 3.2 million government K-12 teachers, my best guess is that 50K positions went unfilled due to lack of money to pay, so that average government K-12 class size went from 15.7 to 15.9.