Unemployment rate

Firstly a good (and plain!) interpretation of the recently-published US unemployment rate:

http://norris.blogs.nytimes.com/2009/08/07/not-as-bad-but-not-good/?hp

Key points:
*that is statistically unimportant given the sampling error in the household survey.

*In any case, it fell not because more people said they had jobs — employment was down in that survey — but because fewer people were still looking for work .

*the auto manufacturing business had added 28,200 workers. Added? That sure is not the impression you’d get from the reports coming from Detroit. ....It turns out those are seasonally adjusted numbers. .....

*The number of unemployed people who have been unemployed for 14 weeks or less was .... the lowest figure for that group since December. But the number unemployed for 15 weeks or more was 7.88 million, up 74 percent since December and the highest figure ever. ....more than a third of the unemployed have been out of work for at least 27 weeks. The average unemployed person had been jobless for less than 20 weeks at the end of last year. Now the figure is over 25 weeks.

*Is it good news that fewer people are losing their jobs? Yes. Is it bad news that the number of long-term unemployed is rising? Yes.

-------------------------------------------------------------------

To have a better understanding, one should remind how unemployment is defined and how unemployment rate is calculated, theoretically.

A very simple symbol ("u" if i m not mistaken) turns out to be pretty complicated empirically. The unemployment rate is collected through CPS, one of the 3 leading labor economics databases (US) for the reasons briefly explained on BLS website.

http://www.bls.gov/cps/cps_htgm.htm#where

How is this work/project done?

http://www.bls.census.gov/cps/bdacodes.htm

The evolution:

http://www.bls.census.gov/cps/bhistory.htm

The questionairres can be found @

http://www.bls.census.gov/cps/bqestair.htm---not readable!

http://www.bls.census.gov/cps/intmanc4.htm

The sampling and weighting---**still pretty conceptual in some senses, but that's enough for most usages.

http://www.bls.census.gov/cps/bsampwgt.htm

The imputation--**N/A

The seasonal-adjusted algorithm---**N/A

Some people sniffed at this release. To this, my response is:Any data has room for politics but one should have faith in the quality of US data.

Data has room for "politics" because the ground/field work has so many to tackle, much more than what is in the theorists' minds. Like many other data, the (US) government never disclose how they finally get that final percentage (to the tenth). But overall, people tend to believe the US and UK data are credible, at least after the adjustments. (what in the news are normally "early release", which means tentative, the "finalized figures" come month(s) later) It probably relates to the tradition, the custom. It is interesting to see how vast the difference is between US, UK, Germany data and Italy and Spain's. Whether it is a well-functioned democracy matters too. Who believes the Chinese data on employment and growth rate?