A client asks why I did not choose to calculate inflation rates excluding apparel prices this month, when doing my usual slicing and dicing of the PCE Price Index. The reason is that I like to have a standard presentation, so that I do not get sucked into thesis creep. It might be good discipline to avoid stripping out this or that as one’s own priors might incline one to do.
However, the risk of data mining on my behalf is reduced if I am just innocently answering a question. And besides, the client raises a relevant measurement issue, because lower apparel prices last month restrained measured inflation more than used motor vehicle prices lifted it. Apparel prices fell only half as much as used motor vehicle prices rose, 1.4% vs 2.8%. But the weight of apparel in the PCE price index is more than twice that of used motor vehicles, so they dominated anyway. (Within the CPI, used motor vehicles have a misleadingly higher weight.)
The table below expands my usual slicing and dicing to include an extra “special factor” that I can switch up each month, depending on what is going on, and trying to be careful to avoid thesis creep. In the table and charts I will just label it “special factor” so that – in haste – I don’t mislabel something. This month the special factor is apparel. I think the table speaks for itself, and I do apologize for it having so many line items. But I have bolded the line item towards which I want to draw your attention.
The Observed Rent Core MPO PCE Price Index Excluding Used Motor Vehicles and the Special Factor (this month apparel) was up 17 basis points on the month. That is very closely in line with the 3-, 6- and 12-month rates annualized, all of which in turn are close to the Fed’s 2% inflation objective. So, does that mean that the earlier inflation overshoot is fixed, even though this specific tweak results in a slightly higher inflation rate this month than in my typical presentation?
No, and for three reasons:
- Market Price Only (MPO) versions of the inflation rate exclude volatile non-market prices, including those for many financial services, which have been a particular source of noise and indeed chronic distortion during the past several quarters. However, we need to recognize that non-market prices tend to rise more quickly than the general price level over time, so stripping them out generates a bias as well as removing noise. Accordingly, a 2% measured inflation rate there would be consistent with underlying inflation of about 2.2% or even 2 ¼%, although with some estimation error around that bias adjustment too!
- Compounding the issue above, Observed Rent measures of PCE inflation replace my estimate of sequential marginal rent growth with the trailing government measures of average rent growth. As I have been belaboring for years now, the Observed Rent measures are probably more representative of current inflation pressures than are the standard price indexes, including the standard Core PCE Price Index. And Fed Chair Powell has been increasingly adamant in recent months that he tends to agree with me (implicitly) on this point. Having said that, though, these Observed Rent measures spot 100% of the benefit of the doubt to my own approach. Maybe, the dispassionate observer, pardon the pun, would assign a 2/3 weight to marginal rents and a 1/3 weight to those government measures of average rents that my own preferred index fully excludes. That dispassionate observer would see an inflation rate 13 bps higher than I tend to. That number is smallish because the gap between marginal and average rent inflation has narrowed in recent quarters, making the stakes lower, and because I am discussing the difference between 2/3 accepting my approach and fully accepting it. But even 13 bps is incremental to other stuff going on.
- Within even my most preferred measure of underlying PCE inflation, the contribution from services inflation is somewhat high in comparison with the contribution from underlying goods price inflation. While services inflation is clearly high relative to what we would assume would be consistent with the Fed hitting its overall inflation rate over time, goods price inflation is arguably running a bit low. It is comparable with where it was prior to the Covid shock, but keep in mind that that was a period of below-target overall inflation. This is relevant to the extent that we might want to overweight the contribution from services inflation on the grounds that it is more persistent and more closely linked to the state of the business cycle, including niggling wage pressures. In making this point, I am not reminding you that services have a higher weight in the overall price index. That is trivial. I am going beyond that to say that prices there should be more highly weighted still. We should slightly overweight their measured contribution.
In case the above is a bit tough to follow, let me clarify by admitting that I manage to be the three handed economist here today. On the one hand, excluding apparel prices is reasonable and does raise the inflation rate meaningfully for October taken in isolation. On the other hand, the single best measure of underlying goods and services price inflation, even taking account of this effect, looked okish in October. But on the freakish third hand, and stepping back a bit to wave it, there is still some wood to chop on the inflation side. Powell is right when he suggests that he has not quite finished the job.
Practically speaking, this probably means that Powell does not want above-trend demand growth in the current environment. And he will be somewhat reactive if we get a tariff impetus on top of the current dynamic. This last point is not new, obviously. And following the advice of the client would – at the margin – make it just a tiny bit more so.
One final thought, I am willing to share the very involved spreadsheets involved in making these points. Just ask sales or me directly. The proprietary part of this stuff is not my calculation skills. Believe me.
Source: BEA, FH calculations, including chaining
Data are actual to October.
