Thursday, March 18, 2010
Well, it looks like whatever it was that was going on last month within the shelter category of the Labor Department's consumer price data is now back to normal in this month's report. As shown below, after considering the respective component weightings, the total seems to make sense in February (in green) versus the oddity that was January (in red).
Recall that, about a month ago in A math problem at the Labor Department? this question of addition was raised after the -0.5 percent decline in shelter costs caused the widely publicized first negative reading on month-to-month core inflation in many, many years.
What looks to be an obvious error was attributed to seasonal adjustment. That is, after the weightings (in blue) are adjusted for seasonal factors they somehow allow the lodging away from home component to greatly impact the shelter total.
All else being equal, you'd have to increase the lodging away from home weighting by a factor of ten to get the overall shelter total as reported in January. Do that many more people travel in January than during other months of the year?