The median CPI, which excludes all the large changes in either direction and is better predicted by labor market slack, is out and is extremely ugly. A 9.2% annual rate in August, the single highest monthly print in their dataset which starts in 1983 (second highest was in June).
(And to address the inevitable objection: median is often just shelter. But median ex shelter is usually about the same as median with shelter. Just like if the median test score in my class was Sarah then the median excluding Sarah is going to be very little changed.)
Trimmed mean, which I think is a slightly less good measure, is also showing a steady upward climb--but not quite as ugly.
Here are the three-month changes for several of the measures: you can see median and trimmed mean firming up even as core (excluding food and energy) comes down.
You can find the data here. And my earlier thread on the BLS numbers here.
clevelandfed.org
clevelandfed.org
And here is my discussion of the Ball et al paper which was part of what prompted me to take a harder look at median CPI and it's benefits as a more pure measure of labor market tightness--less affected by idiosyncratic factors.
Loading suggestions...