Hat tip to Mark Thoma for covering yesterday's speech by Federal Reserve Bank of St. Louis president Bill Poole. Excerpts from the speech can be found in Mark's post, but here is one side comment that is kind of interesting in light of last week's blogland discussions of seasonal adjustment and such:
To emphasize how important it is to examine statistical reports closely, note that even revisions must be examined closely. Consider the April employment report. Seasonally adjusted employment for March was revised up by 93,000, a significant number of jobs. But the seasonally unadjusted count was revised up by only 50,000 jobs—still significant but many fewer than 93,000. Thus, revision of seasonal factors accounted for 43,000 of the total revision of 93,000 in the seasonally adjusted job count. Revisions of seasonal factors are not fundamental, but merely change how employment gains are distributed across the months of the year.
The lesson, by the way, is not that the data are misleading but instead that they must be interpreted with due regard to the statistical processes behind their construction. Current estimates published by the statistical agencies represent the best estimates our first-class professional statisticians can supply. Agency publications provide ample information on the statistical properties of these estimates, and we ignore those properties at our peril.