>Doug:
>> Dunno where the 1.4m figure comes from. According to the BLS's survey
>> of households, employment has risen by 1.1m over the last year.
>> According to the survey of employers (the "establishment" survey),
>> employment has declined by 35,000. According to the BLS, the
>> establishment survey is more reliable (larger sample than the HH
>> survey, and there are great difficulties involved in expanding the HH
>> sample into national figures, because of changes in the population
>> between decennial censuses). Normally the difference between the two
>> surveys is pretty small, but the discrepancies over the last year or
>> two have been unusually wide.
>
>You can also check the data from the Quarterly Census of Employment and
>Wages (aka ES-202) that produces near-census quality of employment and
>wages covered by the unemployment Insurance (UI). The last year they
>published on line is 2002, but 2003 is coming soon.
The BLS's annual benchmark revisions are based on the ES-202 data. (The monthly sample covers 300,000 establishments, about a third of the total employment universe; the ES-202's are near-100%.) The latest benchmark revision, released on Feb 4, revised downward the estimate of March 2003 employment by 122,000 (see <http://www.bls.gov/web/cesspec.htm> for details). Rightwingers, who want stronger numbers, have been touting the HH survey as telling the "real" story, but the benchmark revision counsels against that (though it's rather old news by now).
Doug