On Dec 13, 2008, at 12:24 AM, Patrick Bond wrote:
> My imagination? Here it says there were two substantial series
> breaks (1990, 1994): http://actrav.itcilo.org/actrav-english/telearn/global/ilo/seura/usunemp.htm
The break in 1990 is the result of the application of the decennial census data to the population controls used to inflate the monthly sample of 60,000 households into a national estimate. It happened again in 2000. The BLS reports estimates of the effect of the transition, and the effect is pretty minimal (on the order of 0.1 point).
Of the 1994 redesign, we read at your source: "Beginning in 1994, data are not strictly comparable with data for prior years because of the introduction of a major redesign of the labor force survey questionnaire and collection methodology. Bureau research suggests that the effect of the redesign was to raise the overall unemployment rate by about 0.1 percentage point."
That's *raise*, not lower, and by all of 0.1 point.
Doug