[kenberg]

The Post had a short article today describing how these numbers are obtained and discussing security. I am not sure it can be viewed by non-subscribers, but here is the link.

http://thewashington...per/viewer.aspx

[phil_20686]

kenrexford, on 2012-October-05, 07:48, said:

I read an AP story today discussing the amazing drop (before the election) of the unemployment rate to 7.8%!!! Wow, that sounds good.

Except, I am having truoble with the numbers. The BLS says we added 114,000 jobs in September. The unemployment rate is based upon surveys, and this month's survey indicated that the number of people claiming to be employed soared by 873,000. (The adjustments to prior month's jobs numbers would not matter, because the surveys back then would have accounted for the real situation, not the BLS being supposedly off. Plus, that was only up about 67,000.)

So there are lots of possible explanations.

(1) Unemployed+employed =/= the labour force. The unemployment rate can drop, if, for example, a bunch of unemployed people decide to go to university. Also, with an aging population one expects participation rates to decline, and a larger percentage of your population to be retired.

(2) Most of these surveys based on changes contain systemic errors, so every six months they `rebalance' the survey. I know that this month the BLs `found' 380000 jobs on their (bi-?)annual rebalancing.

(3) The three employment numbers have different lag times, in particular, the survey numbers are normally ahead of the payroll numbers, as many companies seem to do their payroll numbers at least one month behind. The 114,000 number is from the payroll number, so if there has been a sudden improvement in labour markets, you would expect the other numbers to be ahead of the payroll number, even though, in the long term, it is the most accurate measure.