It is important – and I say it every year – to remember that when we are looking at economic data from December (and in many data series such as Employment, January as well) there are massive error bars around the numbers. The government doesn’t report error bars, but they should. Frankly, when it comes to , I barely glance at the number because it just doesn’t mean very much.
The problem isn’t so dramatic in at the headline index level, because the main sources of volatility in the index also happen to be the ones that provide all of the seasonal adjustments, so we tend to miss estimates roughly as often in December as in other months. As we go through the numbers today, however, you’ll notice a bunch of things swinging one way after swinging the opposite way last month. That’s the sort of thing that can easily be caused by the placement of Thanksgiving, so you can see reversals from November’s number to December’s. I am not saying that everything in the CPI report today is infected by that effect; just keep it in mind.
Now, while I say the ‘problem’ of seasonal volatility isn’t as bad in CPI at the headline level, recognize that December sees the most-severe seasonal adjustment to the headline figure. Here are the seasonal adjustment factors for 2023 (they don’t change much). A number below 1.0 means that the seasonally-adjusted headline number will be higher than the nonseasonally adjusted number, because the seasonal pattern ‘expects’ the weakness, and vice-versa. You can see that December is the month furthest from 1.0. What you can’t see from this chart is that if you want to get technical about it, December is also the only month for which we could really reject the null hypothesis that the adjustment factor is 1.0…in other words, the only month where we are really confident that the effect is to cause the NSA CPI to be lower than the average month. November, maybe.
As an aside, this is why April maturity TIPS tend to have higher yields than January maturities. The January TIPS mature to an index that is an average of October and November CPIs, while April TIPS mature to an index that is an average of January and February CPIs. So April TIPS always get an extra December CPI in them, and if there’s one month you don’t want, it’s December. So April TIPS have to have a slightly higher yield to entice people to hold them.
Right, that’s a big prelude discussion. Summing up: don’t get too excited either way with this number. More important is that the overall market has been selling off. 10-year breakevens have risen 14bps, and 10-year real yields have gone up 26bps. How much of this is because of a fear that inflation is turning, is unclear. But in December, the overall data was pretty close to expectations. Core inflation came in at +0.225% m/m, compared to expectations of +0.25%, which is less dramatic than it looked when rounded and it printed at 0.2% vs expectations of 0.3%. A small miss lower, and to be fair the best core number since July.
Headline was only 0.04% NSA…which gets adjusted to +0.39% when the seasonally-adjusted number is reported. See what I mean? So we look at the y/y numbers, which basically replaces last December with this December (thus neatly avoiding the seasonality issue). Y/Y headline CPI rose from 2.73% to 2.90%, and Y/Y core fell to 3.25% from 3.30%.
You may notice that none of those numbers looks like it’s at 2%. Nor is Median CPI, which was (my estimate) +0.31% m/m, the highest since September. If I’m right about that print then the y/y would drop to 3.86% from 3.89%.
So on the macro side, top-down, this does not look like the sort of data that the Fed was expecting when it started easing in September. Since in my opinion this has been eminently foreseeable for a long time when you looked at what was driving CPI, the conclusion must be either that the Fed is just incompetent when it comes to inflation forecasts, or it doesn’t care about inflation, or the rate cut had nothing to do with economics and was just a political gambit to get Harris elected. None of those answers is flattering. I suspect answers #1 and #3 are the main drivers of the most-recent policy error.
The good news in the inflation figures is that there’s no one major group that still looks alarming.
When we drill down to the monthly data this month…that’s where you see the seasonal volatility. For example:
- Used Cars was expected to be roughly flat. It was +1.2% after +2.0%.
- Rents rebounded; OER and Primary Rents were +0.31% after +0.23% and +0.21% respectively last month.
- Lodging Away from Home was -0.95% this month; it was +3.16% last month.
- Airfares were +3.93% this month; they were +0.37% last month.
- Car and truck rental +0.58% this month; -2.99% last month.
- Baby Food +0.42% this month; -0.12% last month.
- Medicinal Drugs +0.08% this month; -0.10% last month.
- Doctors’ Services was lower, +0.06% vs +0.28% in November; but Hospital Services were higher at +0.23% compared to 0.00%.
A few broader observations. Core Goods and Core Services both continue to move back towards zero: goods from underneath and services from above. CPI for Used Cars is still -3.4% y/y, and I’d expect it to slowly recover from the spike and reversal stemming from COVID. But we now have an extra factor, and that’s the devastating California wildfires. There are two things you see burned out in every picture. Vehicles, for one. Used and New car inflation is going to turn higher, and maybe quite a bit, going forward as people in California need to replace their wheels. Over the medium term, the dollar’s strength would help keep core goods inflation tame and even slightly negative, but thanks to the wildfires we are likely to see core goods back above zero shortly.
And the other thing you see burned out, of course, are houses. Primary Rents have been slowly converging with our model, but rents are going to get goosed in California immediately and that effect will be smeared out because of local laws against ‘price gouging’ that prevent landlords from hiking their rents immediately to the equilibrium level implied by lots more demand and lots less supply. So they’ll hike, but it will take longer. This is mainly a California effect, naturally, but it will be large enough to affect the national numbers.
Incidentally, you’ll also see these in Lodging Away from Home inflation not just in California but in the entire western US. And maybe further, since remote work makes it possible to temporarily relocate almost anywhere. Federal support of the displaced will ensure that is not a 1-month effect. So in shelter, January and February (and beyond) numbers are going to be a lot more important than today’s release.
I am sure that will be used later to argue that “this inflation in 2025 is all due to the wildfires,” but we should remember that inflation in 2024 was (at best) leveling out and possibly hooking higher again. Broad core inflation ex-shelter has now risen four months in a row. It isn’t alarming, at 2.12%, but it isn’t just shelter keeping inflation above target and the story in early 2025 won’t be ‘all about shelter and cars.’ Supercore is also improving, but it isn’t going to pull the overall CPI down to target if Shelter doesn’t keep decelerating and as Core Goods goes back positive.
Supercore is indeed looking better, but we still have wages rising at 4.3% y/y. Remember that wages and supercore are modestly cointegrated. Or, in English, supercore is where wage-driven inflation tends to live. Wage growth needs to soften a lot more in order to get supercore back to target-like levels.
Again, all of this is December and in January we have had a massive natural disaster that will affect inflation data as soon as next month – and for months going forward. This will obfuscate the fact that the Fed already made a second policy error (after the COVID-era error of adding too much liquidity and not pulling it back quickly enough), dropping rates prematurely and letting money growth re-accelerate (M2 y/y is at 3.7%, but annualizing at 4.7% over the last 6 months and 5.8% over the last quarter ended in November). The bottom line is that the December inflation data is just not very important. What happens next…and what is already happening…is the story that will drive inflation and markets in 2025.