Sept. CPI came in at 0.4% headline and 0.3% on a core basis (excluding food and energy), both month-over-month, seasonally adjusted. The 0.3% core month-over-month increase was slightly worse than expectations (most people were expecting somewhere north of 0.2%, but after that it becomes a rounding debate...this month's unrounded figure was 0.27% vs. 0.28% last month). Year-over-year, CPI including food and energy increased 3.7%, while "core" CPI (the more important figure) increased by 4.1%.
Under the hood though, there's definitely some things to pick at, some of which look like they could be an emerging problematic trend, others perhaps not so much.
As usual, let's look at the CPI through our three different "buckets": goods, services ex. shelter, and shelter. For September, goods contributed -11 bps this month, services ex. rent contributed +12, and shelter contributed +26.
All three buckets had things to call out. First, in goods, 9 of the 11 bps negative contribution came from reductions in Used Car prices. New car prices increased for the month, contributing +2 of the -11 bps. These compare to the pre-COVID average contributions of essentially zero for both categories. This tailwind (meaning something that produces lower inflation) probably won't be quite so robust going forward, as used car prices seem to have shallowed out and even slightly increased in September (see here: https://site.manheim.com/en/services/consulting/used-vehicle-value-index.html).
It's possible this firming in the market has come from dealers looking to bolster used car inventories ahead of the UAW strike, but that remains to be seen. Given car prices can swing the CPI (and certainly have in recent years), this category could be bumpy over the next few months, both because of the strike, but also because the significant move up in interest rates could put more pressure on prices again. As a reminder, CPI tends to trail third-party data indices like Manheim by at least a month, and sometimes more. Bottom line, a glass-half-full interpretation would be that goods ex. cars was still negative, slightly better than pre-COVID averages. A glass-half-empty view would be that this 9 bps tailwind from used car prices won't last, and thus, "run-rate" inflation for the month was actually more like +37 bps (4.4% annualized) than the 0.27% (3.2% annualized) we actually reported.
Second, the services-ex-shelter category contributed +12 bps this month. This was the second month in a row where this figure was above the pre-COVID average (+7 bps), though lower than last month's +17 bps (which we said was uniquely boosted because of one category, transportation services). Contrary to last month, the contributions to services ex-rent category were more broad based, which isn't exactly what we want to see (ideally we'd see inflationary breadth continuing to slow down, like it did in this category for most of the last year). As we've seen, the CPI and its various components can very much ebb and flow, and nothing in the CPI is a ever straight line up or down. Since this category is a focus at the Fed, however, this month's news isn't particularly helpful.
Lastly, we have shelter, which contributed +26 bps this month. This was the biggest oddity for the month, as it was the largest contribution from this category since May, when shelter also contributed +26 bps. In recent months, this contribution has been coming down, which was expected given 1) the lag in CPI data 2) home prices and rent prices remain under pressure or at worst, are modestly inflationary. The pre-COVID average here, as a reminder, was +11 bps a month.
As a reminder, the CPI calculates shelter inflation in a bit of a strange way. The easiest way to explain this is to look at an example where you have 12 people living in an apartment complex, and each person signed their lease in each of the last twelve months. Let's say I'm the person that signed their lease that started last October, and let's also say I just renewed my lease Oct. 1, 2023. Let's further say my rent on the year rose by 1%. The way the CPI counts shelter inflation, they will not treat this 1% as the new market rate for the whole building. Instead, they'll continue to count the monthly increases for all the other tenants who renewed in the last twelve months in their calculation for this month's shelter inflation. The effect of this is that it takes time for changes in market prices for shelter to flow through the CPI's shelter index (at least 6-12 months), simply because it takes time for the other tenants in my building to renew their leases at this lower +1% rate. Market participants and forecasters have given greater emphasis to market-based rent indices, which continue to show that rents are actually going down (so rents are de-flationary, not just -dis-inflationary). See here: https://www.realtor.com/research/august-2023-rent/,
here: https://www.apartmentlist.com/research/national-rent-data and here: https://www.costargroup.com/press-room/2023/apartmentscom-releases-rent-growth-report-third-quarter-2023 for evidence of this. Home prices have started increasing again, though modestly (see here: https://www.nar.realtor/blogs/economists-outlook/august-2023-existing-home-sales-fall-as-home-prices-increase-with-mortgage-rates#:~:text=NAR%20released%20a%20summary%20of,by%2015.3%25%20from%20August%202022.) but it's very unclear if this will continue given the significant rise in interest rates over the last month, which has pushed mortgage rates approaching 8%. Regardless, it's a little strange why the CPI shelter index ticked up so much this month, but reason stands to indicate the contribution will come down in upcoming months, more in-line with the trends seen in June, July and August.
Away from our three baskets, let's look at category level changes to get a broader feel for inflationary breadth. This too was not great this month, indicating the worst inflationary breadth since January of this year, though this was almost entirely confined to the "core services" category. As a reminder, we like to look at what I call "Sub-Indent" 4 and 5 in each month's Table 2 of the CPI release. Sub-Indent 4 is a basket of 55 goods and services categories that combined make up about 98% of the CPI, while Sub-Indent 5 is a basket of 101 goods and services categories that make up about 78% of the CPI. We can also call Sub-Indent 4 and 5 "Level" 4 and 5 to make it easier, so let's just do that.
This month, the median increases for Level 4 and Level 5 categories was 0.3% (30 bps) and 0.2% (20 bps). The average increases were 0.19% for Level 4 and 0.23% Level 5, but median is usually the better metric. These charts are below, along with Level 4 Category charts broken down into goods, services and services ex. rent as well. They tell the same story we told above, which is that September was not a great month for inflationary breadth, though this was almost entirely in the services ex. rent ("core services") category (the typical goods basket in Level 4 categories declined slightly this month, by -4 bps, so essentially flat, and in-line with pre-COVID levels, similar to what the aggregate goods basket did as well).
Now here's our 3 month rolling average of Level 4 and 5 category level inflation versus reported core CPI. As you can see, category level inflation remains below the aggregate (core) index, suggesting positive trends for inflationary breadth, but this month took the average in the wrong direction.
Bottom line, September CPI had its oddities, but on balance, was a "glass-half-empty" month, and given the recent rise in interest rates, isn't going to be helpful in getting the Fed comfortable to bring them down anytime soon. As we've cautioned in prior months, one-month does not make a trend, but given some of these items have now been occurring for two months, we need to keep our eye on the problem children going forward. The tailwind from health insurance is likely to go away soon (which has been 2-3 bps a month), so if shelter continues to be slow in coming down, the used car price contribution goes away, and services away from shelter remain hot, we could see some uglier CPI's in coming months.
There's still plenty of evidence to suggest that inflation is coming down, and that we're probably much closer to the Fed's 2% target than most people think, but being intellectually honest forces you to admit that this month's CPI release isn't one of those datapoints.
Comments