I was mulling this morning’s sucky employment report (which sucks, don’t get me wrong) and was curious as to the effects of the seasonal adjustment process. You might expect a seasonal increase in non-farm payrolls for November from late holiday-season hiring, so it would make sense for the seasonal adjustment to knock the NSA figure down. But how much? Here’s the last three years of October-November employment change:
- November 2008, NSA NFP -685K, SA NFP -728K, net adjustment -43K
- November 2009, NSA NFP +80K, SA NFP +64K, net adjustment -16K
- November 2010 (preliminary), NSA NFP +217K, SA NFP +39K, net adjustment -178K
There is, of course, no law of econometrics violated here. The adjustment process is complex and doesn’t assume the effect is the same every year (a feature, not a bug). For instance, the not-well-loved net birth/death adjustment is applied before seasonal adjustment and a note on the BLS website regarding the effect of the net birth/death model on seasonally adjusted payrolls basically says, “Don’t ask because we can’t tell you.” So. But still.