Doug Henwood wrote, >More vulgarly empiricist statistics from the reeking stinkpot of bourgeois >mystification, the BLS.... It isn't the vulgarity or ideological taint of BLS statistics that makes them inappropriate for the kind of sweeping inferences that Doug sometimes draws from them. It is the AGGREGATE nature of the particular time series. BLS produces wonderful statistics, as does Stats Canada, but you have to be careful about how they're put together. So when Doug asks, >By all accounts, job market volatility in the U.S. is higher than it was >10, 20, 30 years ago. Then why is the "experience of unemployment" series >behaving this way? The answer may not be at all evident in the series. Over the past 40 years there have been tremendous shifts in the sectoral composition of the labour force (namely from manufacturing to services), its demographic composition (higher proportion of women; digestion of the baby boom), and the rate of unionization -- just to mention three biggies. The experience of unemployment series lumps together the experience of production line workers and government employees, etc. etc. and thus the average it expresses is not terribly meaningful in reflecting _job market_ volatility. To vulgarize dual labour market theory, let's just say there's a job market and a job non-market, as the non-market increases in proportion and becomes more sclerotic, the market contracts and becomes more volatile. The aggregate effect, "experience of unemployment", would decline. Doug, try this exercise: put one foot in a bucket of ice and the other foot in a pot of boiling water. Describe your comfort level. Regards, Tom Walker ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ knoW Ware Communications | Vancouver, B.C., CANADA | "Only in mediocre art [EMAIL PROTECTED] | does life unfold as fate." (604) 669-3286 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ The TimeWork Web: http://mindlink.net/knowware/worksite.htm