November CPI Comes In A Lot Cooler Than Expected, But There Are Issues...

October CPI comes in...not at all.

The BLS released the Consumer Price Index for November this morning and it was much cooler than expected. It was also uniquely weird: due to the gov’t shutdown, data for the October CPI was not collected, meaning we don’t have the usual monthly changes.

The table below shows the topline numbers along with expectations. The CPI rose 2.7% over the past year, down from September’s 3% and the same rate as July. But the yearly core rate came in even lower, 2.6%. the lowest since March 2021.

The table also shows the two-month changes: both indexes rose 0.2% from Sep-Nov change, or 0.1%/month if you want to take an average.

Weird, I know, to have a hole in the data for Oct. In a series that began in the late 1940s, before even I was born!, the Bureau has never missed a month. It’s a terrible statement about gov’t dysfunction, but it also makes it hard to parse the current inflation situation.

Omair Sharif is one of the most careful inflation analysts on the beat and he identified one problem with the November CPI data which helps explain the expectations’ miss:

The BLS just assumed [shelter costs] were zero for October. I am sure they have a good technical explanation for this, but the only way you get a two-month average for rent of 0.06% and OER at 0.135% is assuming October was zero.

And, remember, “shelter,” which normal people call “housing,” is 35% of the overall index and an even larger share of the core. Here’s a little table I made wherein a backed out the Bureau’s Oct assumption by taking a mean of the adjacent months of Sep and Nov. Not exactly zero, but close, and likely too low.

Bottom line, I wouldn’t try to make too much out of these numbers, especially monthly changes. And if the Oct change is biased down, that takes a nibble out of the yearly rate too, because it means the November rate is coming off a baised-down base. Also, while data collection for Oct was just about non-existent, it was also incomplete for November, with many more observations from the latter half of the month, thereby possibly overweighting Black Friday sales.

Can we learn anything from the report? Sure, in that big price changes over the past year will not be totally obfuscated by one-month’s missing data. Eggs are down 13% over the past year, but beef is up 15% and coffee is up 19%. I’ve been carefully tracking core commodities as that’s where you see tariff impacts, and it’s been trending up for awhile (though the increase starts pre-tariffs). The dip in October, using the same adjacent-mean process as above, should be treated with caution.

Bottom line, perhaps there’s less stagflation than we thought, and more just a weak labor market and easing inflation. But you shouldn’t update your priors based on one outlier report, and that’s especially the case with this unusual report.

More to come as I continue trying to see through the data fog!