CPI for January 2022

We've been working with the CPI data since last week, and noting the steady increase in the CPI inflation rate that's been going on for 14 months or so.

First week of the month gets us labor market data (from the BLS), the second week of the month gets us price level data. 

The BLS press release shows inflation (MoM) of 0.6%,  and 7.5% (YoY). The latter is an increase. But ...

Do note that the BLS does seasonal adjustments on a fairly regular schedule, but that they also do a big one for the whole previous year that comes out in the February report. 

They also do seasonal adjustments across years. Think about it, if right now they modify January to December of last year, but January (as announced a year ago) was building off the just previously revised year. Since they're tweaking this year a little, they also tweak that year a little, and so on, to eliminate any quirks they may have induced in the data. In principle, as they carry this back further and further, the modifications needed get smaller and smaller, and they are not done at all more than 5 years back.

Bottom line is that we should download the new data, and duplicate our earlier analysis with the new numbers.

***

Also, while I am grousing this month about issues with BLS deseasonalization, please keep in mind that deseasonalization is a good thing. If they do it, it's helpful. If they don't do it, you'll need to. And while criticism is merited sometimes, it's not clear that you'll do a better job than they do. What is important is that you can explain what you did to others (and they can debate it), whereas with "official" deseasonalization, we're often dealing with a black box.

Another factor to consider is that an agency like the BLS is deseasonalizing all their data jointly. So an odd adjustment in one spot is at least reconciled with all the data that relates to it. This is a big plus that's hard to duplicate working on your own with a handful of series.

***

One bit of weirdness is that the BLS makes a point of noting that the 7.5% YoY figure is based on not seasonally adjusted data. However, as I argued in class, forming a YoY is a form of seasonal adjustment; and if you're logging the levels of the data, as a difference over 12 months, it's actually a preferable form of seasonal adjustment.