- by New Deal democrat
In my note yesterday about the July CPI, I noted the transition from the trend where overall inflation ex-shelter was low, and shelter was high but disinflating, to a trend where inflation ex-shelter while still low was increasing, as shelter contributed the most to disinflation. The question was, and will be going forward, how much are tariffs contributing to inflation?
In the past 24 hours I’ve seen a number of “just-so” indexes; basically, if you exclude things that aren’t going up (e.g., gas and new cars), everything else is going up! Well, duh! So I am unconvinced by those analyses.
Let me start by re-upping two of my graphs from yesterday. First, headline inflation vs. core vs. all items less shelter:
All items less shelter have been increasing at less than 2.5% for two full years. Since officially measured shelter costs have been decelerating all during that time:
the headline number has decelerated as well. But inflation ex-shelter seems to be trending higher in the past 9 months, meaning core inflation has already stopped decelerating.
So let’s divide up everything *except* shelter, which oddly enough is counted among the “services” sector of the CPI. The below graph shows the YoY% change in consumers costs of durable goods (gold), non-durable goods (red), and services excluding shelter (blue):
The cost of durable goods had actually been undergoing *deflation* during most of 2023 and 2024, but the trend has reversed higher. The inflation rate for services has also ticked up mildly in the past few months, while that for non-durable goods has been meandering around 1%. Below I show the monthly changes in the first two since the beginning of 2023 better to show those trends. I excluded non-durable goods because that would just be a squiggle:
The uptrend in both durable goods and, since last summer, services excluding shelter is apparent.
Supposedly so far producers have only passed on a small portion of tariff costs to consumers, but that won’t last forever, so - IF inflation statistics remain reliable in the coming months - I expect those costs to start showing up particularly in goods prices.
The reliability of BLS statistics going forward is a real concern, given T—-p’s nomination of E. J. Antoni to become Chairman of the BLS. Last year he authored a paper arguing that the US had been in recession since 2022, using a “unique” measure of inflation that used the Housing Affordability Index instead of OER for the shelter component.
The Housing Affordability Index consists of two components: the change in house prices, as well as the change in mortgage rates. In terms of inflation, both of those components have uses. For example, no less than Barry Ritholtz of the Big Picture has argued that Fed rate hikes actually *increase* inflation but raising the costs of mortgages in particular. And I among others have argued that replacing OER with House Cost Indexes is a better model for consumer prices (bearing in mind that in any given month or year only a small fraction of consumers make a new house purchase). Further, I have argued that because house prices lead OER by 12-18 months, the Fed should use a “House price indexed CPI” in setting rates, since that tells them where inflation is likely to be in a year or so.
But putting those two components together to measure inflation strikes me an another “just-so” compilation, designed to arrive at a desired conclusion, i.e, Biden’s Presidency featured a long recession.
And interestingly, Antoni’s analysis begins in the year 2019. I have learned that politically motivated economic analysis often makes use of cherry-picked start or end dates - and it turns out that this is exactly such a case.
Below is a graph of house prices as measured by the FHFA, together with the Fed funds rate over the past 10 years. Remember, it is the combined effect of these that makes up the Affordability Index. And lo and behold, look at what happened in 2017 and 2018:
House prices went up 15%, and mortgage rates increased about 15% as well, from 4.20% to 4.87%.
In other words, it appears that Antoni’s own analysis would also show a recession during the entire first 2 years of T—-p’s first term. Ooops!
