Financial Post: "Most annual measures, notably the annual CPI used to index the tax-and-transfer system, compare all 12 months of one year with all 12 months of another. But the labour force survey year-end review insists on comparing December this year to December last year, which is not only confusing but also vulnerable to distortions from an outlier in either December. Pity the poor reader of StatCan press releases trying to grasp what the agency means by “a year.”
"Some differences in what appear to be the same variable are inevitable and cannot be eliminated. The widely-cited payroll data on jobs (the preferred measure in the U.S.) is not and never will be the same as employment derived from the survey of households. Payroll data do not cover the millions of self-employed workers, while people who moonlight in more than one job will appear only once in the household count of employment though they register on more than one payroll. Some statistics are necessarily complicated. But statistical agencies should try to minimize complication, not gratuitously add to it simply to satisfy the whims of division heads."
Comentários