August CPI increased 0.2% and 0.3% on a headline and core basis, respectively, month-over-month (and seasonally adjusted). This was about one-tenth (0.1%) hotter than most forecasts in both cases, but as we'll discuss more in a minute, the key driver of the "miss" versus expectations was due to the item that's been a thorn in our sides for a long time now: shelter.
Unrounded, core CPI increased +28 bps this month (0.28%). As the table below shows, this puts the year-over-year figures at 2.5% for the headline CPI (40 bps below last month's 2.9% and the lowest since February 2021), while the core inflation year-over-year remained at 3.2% (similar to last month). As we've discussed in prior posts, the year-over-year figures are less relevant given "base effects." If we want to know how much inflation is in the economy right now, we need to understand how fast the American economic car is driving now, not how fast we're driving relative to a year ago. This makes the 1, 3, and even 6 month annualized figures better barometers for assessing current inflationary trends than the year-over-year figures.
As we normally do, let's go through our 3 buckets.
First, goods contributed -4 bps to core CPI this month, relatively in-line with recent months but below the pre-COVID average of 0. This is good, and consistent with recent months. There's not much else to see here.
Next we have "core services," or "super core," otherwise known as services ex. shelter costs. Here again, the figures were quite encouraging, though perhaps not quite as encouraging as the last two months. This month core services ex. shelter added +8 bps to core CPI's +28 bps for the month, above last month's +5 bps and also above May and June's -1 bps contribution as well. That being said, this month's +8 bps compares to the pre-COVID average of +7 bps for this bucket, so this is hardly concerning (the data is even less alarming when you consider that auto insurance added +2 bps to overall core inflation this month versus the pre-COVID average of 0, so less bad than recent months but still above where it should eventually revert to). The bottom line here too is that core services inflation, long the focus of the Fed, remains consistent with 2% inflation (which is the Fed's target).
Lastly, we have shelter, which was again the bugaboo of this month's report. Shelter contributed +23 bps this month, the most since January 2024 (+27 bps) and September 2023 before that (+26 bps). This was well above July's +18 bps contribution and almost triple May's +9 bps. We've talked (endlessly) about the lag effects from the methodology from the BLS on this category in the CPI, so we won't do so again here. Who has the energy anymore? The bottom line is it still is printing higher figures than anyone really understands, and nobody knows when the CPI will catch up with the market.
The only thing really left to do then is to demonstrate why the inflation rates for shelter in the CPI are not consistent with what's happening on the ground in the real economy. Here's what the other publicly available sources on rent and shelter data are saying:
The first chart below indicates rents have been pretty flat for the better part of two years now (though with some seasonal fluctuations baked in, though as a reminder, the CPI does adjust for this)
The second chart shows the month-over-month changes in national rents. August's report showed rents fell 0.1% month-over-month (very atypical given many people rent during August, so it usually sees among the seasonally highest increases). This compares to the August CPI's report that stated shelter prices increased 0.4% this month (an annualized rate of almost 5%).
Here's some commentary from the ApartmentList August report: "Since the second half of 2022, the seasonal declines in rent prices that take place during the fall and winter have been steeper than usual and seasonal increases of the spring and summer have been more mild. As a result, apartments are on average slightly cheaper today than they were one year ago. Year-over-year rent growth currently stands at -0.7 percent and has now been in negative territory for over a full year." Thus, CPI figures are telling us rents are increasing not quite as fast, but almost as fast as they were in 2021 and 2022. The ApartmentList.com report stands in pretty stark contrast to that, and as we'll see in a moment, it's not alone.
Now let's look at the Realtor.com rent market data. Notice the black line in the first chart, and all three lines in the second chart. Each has been negative for the better part of a year now, indicating rents are declining year-over-year, not increasing at rates well above pre-pandemic levels.
Additionally, the red line in the chart below (showing rent in dollar terms) corroborates the Apartment List figure above in that excluding seasonal fluctuations, rents have basically gone nowhere for 2 years now.
Notice the purple line (this year's rents) has been below the blue line (last year's rents) for most of this year, and is now directly on top of it. The two lines being on top of each other indicates rents are essentially flat year-over-year.
Zillow's data has routinely been indicating higher rent increases than the ApartmentList.com or the Realtor.com reports, for what it's worth, but those figures have still been below CPI.
So here we have multiple sources telling us that rents on the ground are barely inflating, if they're inflating at all. It's true Zillow's data is higher than the other two, but even Zillow's data is not inflating at even close to alarming levels (shelter inflation historically is higher than other categories in the economy). The lags in the CPI's shelter cost methodology are real, and they are frustrating. We recently wrote about why economists don't use the data coming from the stock market enough to help them forecast the economy (and thus to make economic policy). Instead of focusing on the "hard" data, they focus very intently on government produced data, most of which comes from "soft" surveys.
This notion can equally be said of "high frequency" or "alternative" measures of economic indicators that do not come from the stock market, but rather from other private (but still free publicly available) data providers. The rent market trackers we discussed above are good examples of this. How do I know that the Fed doesn't seem to be giving this real-time, market based indicators much weight in their forecasts? Because they say so.
Take this exchange from Chair Powell's January 2024 press conference:
RACHEL SIEGEL. Hi, Chair Powell. Rachel Siegel from the Washington Post. Thanks for taking our questions. So, over the past few years, there have been all these real-time indicators that helped us gain a sharper understanding of where the economy was, like OpenTable data or office attendance. You’ve talked about vacancies in the past. And I’m wondering, at the start of this year, what might be on that dashboard for you that’s giving you the clearest picture of the economy—including on rents—if you could touch on that.
CHAIR POWELL. Including?
RACHEL SIEGEL. Rent. Rent costs.
CHAIR POWELL. Yeah. Well, so we’re not—you know, it’s not the pandemic, so we can actually rely on more, more traditional forms. People are working. They’re getting wages, and, and the economy has largely reopened and is broadly normalizing, as you see. So I wouldn’t say we’re looking at that, that sort of more innovative data as much. You know, you point to rents. So, of course, we follow the, the components of inflation very carefully. Which would be: Goods inflation—I talked about that a little bit; you mentioned housing inflation. So the question is, when will these lower market rents find their way into measured rents, as measured, measured in PCE inflation? And we think that’s coming, and we know it’s coming. It’s just a question of when and, and how big it’ll be. So—but that’s in, in everyone’s forecast, I would say. So that will—that will help.
The question then becomes, why would you not look at this stuff? Why is it "innovative data" and even if it is "innovative" (hardly a pejorative term), why would we not use data that helps us get to a better answer for this very important question? In a world where every five minutes we see commercials about "Big Data" and how it's going to change the world, it's amazing how much of it we already have yet that we don't use.
The good news is that while the Fed doesn't seem to be looking at alternative data, at least in the case of rent and shelter prices, the Fed is aware of the lag in the CPI data. Of course, in order to be aware of the lag though, one has to be aware of the "innovative" datasets we mentioned above that produce the lag in the first place. It therefore doesn't seem quite as likely that the Fed will let interest rate policy be affected by shelter prices if both goods and services remain consistent with 2% inflation (as they are now), even if shelter costs continue to lag. But who knows. This is still at least up for debate and remains to be seen how they'll proceed.
In closing, there's a couple other stats worth highlighting this month.
As we do each month, let's look at category level inflation and see what that tells us. As a reminder, we do this to get a better sense of inflationary breadth. If a lot of categories of goods and services are inflating at rates above 2%, then the true "run rate" of inflation in the economy is probably above 2%. If more buckets of goods and services are running at or below 2%, we can get more confidence that things like shelter (which have a greater weight in the aggregate CPI index) are distorting overall (and aggregate) inflation levels.
For August, the median and average increases for goods and services in Category Level 4 (which consists of 77 baskets of goods and services and comprises 98% of the core CPI) and Category Level 5 (101 baskets comprising 78%) was 0.0% and -0.03% for Category 4, and 0.10% and 0.18% for category 5. Each of these figures, annualized, would produce inflation at or below the Fed's 2% target. Inflationary breadth is subdued, and consistent with 2% inflation.
Notice in the charts below how the lines are below the bars. This indicates that category level average and median increases are below the increases for the Core CPI index as a whole. This tell us that items with a large weighting (like shelter) continue to drive up the aggregate level of inflation, rather than the index being elevated because many different categories remain highly inflationary. This is simply no longer the case. All we're waiting for now is shelter, which we know is lagging because of how the BLS calculates it.
Now here's the category level figures shown on a 3 month basis, and annualized. Again notice how far below the bars the lines are. Category level inflation indicates that we're clearly at levels consistent with the Fed's 2% inflation target.
To be clear, no one is arguing here to "exclude shelter" from inflation estimates. Shelter is a large part of the CPI for a reason: it's our largest cost that we pay as consumers. Instead, we're arguing which data should be used to help assess what the inflation rates for shelter actually are in the economy. I argue that we ought to move off the CPI's lagged methodology and be more open minded about using "market based" "innovative" data sets that provide better real-time, accurate on-the-ground figures. I do not understand why we (or the Fed) wouldn't do this.
A smart friend of mine made the point that we're still just making up for the CPI under-reporting shelter increases in 2021 and 2022. That may be so, and using these "market based" shelter cost indices as I advocate would have made inflation even higher in 2022 and 2023 than was reported. This would in theory mean interest rates should have been risen even higher than they actually were. Both things are probably true, but that's in the past. We make policy for the future, based on what we see on the ground right now. Just because our picture of reality was distorted in 2021 and 2022 because of a flawed CPI methodology then doesn't mean we should continue to make policy based on that distortion now. That's all we're arguing here. To say it differently then, we're not arguing to exclude rent inflation. We're arguing about what rent inflation actually is.
See you again soon.
Comments