The Federal Reserve has a goal of getting inflation down to 2%. But of course, there are a lot of different ways of measuring inflation. Many people know about the Consumer Price Index, and the various ways it can be sliced and diced. The Fed, however, focuses on a different index — Personal Consumption Expenditure — which differs from the CPI in a number of ways, both in terms of category weightings and methodological approaches. So why are there different measures of inflation? Why does the Fed prefer PCE? And how is PCE actually assembled? On this episode, we speak with Omair Sharif, founder and president of Inflation Insights, as well as Skanda Amarnath, executive director of Employ America. We explore these two different measures, the approaches for calculating them, and the weird quirks underneath the surface that makes them all so interesting and controversial. This transcript has been lightly edited for clarity.
Key insights from the pod:
The difference between CPI and PCE — 4:43
Why does the Fed focus on PCE? — 6:44
The wedge between CPI and PCE — 8:26
Why the difference between CPI and PCE matters — 13:17
How does the government measure basic prices? — 15:43
The deterioration of survey responses — 18:37
How the government deals with imputed prices — 20:33
Benchmark rates, banks and inflation — 21:54
The treatment of rent and shelter costs — 24:22
Relationship between goods and service prices — 27:56
Seasonality in inflation data — 31:54
Measuring insurance inflation — 35:34
Measuring airfare inflation — 39:02
January PCE predictions — 40:06
---
Joe Weisenthal (00:20):
Hello and welcome to another episode of the Odd Lots podcast. I'm Joe Weisenthal.
Tracy Alloway (00:25):
And I'm Tracy Alloway.
Joe (00:27):
Tracy, you know, it's funny. There's all this talk about when is inflation going to get back down to 2%? When is the Fed going to hit its goals? But when we talk about inflation, there's a million ways to measure that -- or at least two or three big ones.
Tracy (00:41):
Yeah, this is one of the things that I've really come to appreciate over time. There are so many different flavors of inflation, so many different ways of measuring it. My understanding is that there's basically, if you're going to break it down, there's CPI, there's PCE, there's Core and Super Core of each of those. And the way to do it is just choose whichever one of those confirms your priors and then focus on that.
Joe (01:05):
That's what I do.
Tracy (01:06):
That's the secret.
Joe (01:07):
Whether it's PCE, whether it's CPI, I look for the one that's closest to 2%. And if neither are that close, then I lop out something because it’s like ‘Oh, if you exclude rent and used cars and food…’
Tracy (01:20):
Everything you need to survive?
Joe (01:21):
Then we’re at 2%, so it's time to cut rates. That's basically my approach, but I'm glad to hear that validated by you.
Tracy (01:27):
I mean, that is the running joke, isn't it? If you exclude everything that you need to live, then inflation is coming down. But I think the important thing, the really important distinction, is CPI versus PCE.
CPI is probably the one that people have in their heads when they think about inflation. But PCE, of course, is the Fed's preferred measure and the thing that the central bank is actually focusing on.
Joe (01:51):
And I don't know why that is, actually. I know that most of the time when people talk about what’s [the] inflation rate right now, they'll look at CPI or maybe Core CPI. But then we're always like ‘You know the Fed looks at PCE,’ and I guess I believe that to be true.
But I actually don't know what it is about PCE that the Fed prefers. I know there's some different weights related to rent and there's some few things there that cause them to change trajectory from time to time.
Tracy (02:18):
I think it smooths some stuff. Is that right? It's supposed to be less volatile. But one thing I do know, and I have to admit something here and it's kind of embarrassing at this point in my career as a financial journalist, but CPI includes something called Owners’ Equivalent Rent, which is basically a measure of the cost of home ownership. And it comes up all the time as a key difference between CPI and PCE and I for the world do not understand -- well, I kind of get what it's supposed to be -- but I don't get how it's measured at all.
Joe (02:52):
Well, you know what I've been thinking generally? I think it would be good to do more episodes about how we actually get the data that we get? How do they do the Jobs Report survey? I actually don't really know much about that. How do they do all these different surveys?
All the smartest people we talk to tend to know this stuff really well. Anyway, this episode that we're recording right now, when you're listening to it, happens to be PCE day, the day that the Fed's preferred inflation measure comes out. And so we should understand what's in the Fed's preferred inflation measure and how it differs from other measures of inflation.
Tracy (03:29):
I am really into this topic. I feel like this is going to be an episode that I bookmark the transcript of and then go back and look at it over and over again. So I'm looking forward to it.
Joe (03:38):
Well, I am excited to say we do literally have the two perfect guests to talk about this, two people who really have a deep understanding and appreciation for these numbers appear on the screen. PCE for me is just a number that appears at 8:30 AM on my Bloomberg terminal. But the reality is there are all these surveys and calculations and then tabulations. And so there's a lot of hard work that goes into producing this number. So we're going to be talking to two people that have a deep understanding and appreciation for what goes in from a real bottoms up perspective, how these numbers appear on our screens.
We're going to be speaking to multiple-time guests, both of them. Omair Sharif, he's the founder and president of Inflation Insights. As well as Skanda Amarnath, executive director at Employee America. So Omair and Skanda, thank you both for coming back on the show.
Skanda Amarnath (04:28):
Thanks for having us.
Omair Sharif (04:29):
Thanks.
Joe (04:30):
Omair, why don't we start with you? Why don't you answer the sort of basic question for us of what is the difference between PCE and CPI? Why do we even have two separate measures of this?
Omair (04:43):
Yeah, I think the simplest way to think about it is just that they're intended to measure somewhat different things. So they're designed to sort of do different things. The PCE is a much broader index. It captures essentially more of the economy, if you will, than what the CPI does. The CPI focuses a bit more on consumers’ out-of-pocket expenditures. Whereas the PCE covers not just that, but also what is sort of paid on your behalf by third parties or the government.
And a good way to think about this is healthcare. In the CPI, largely it's measured as your out-of-pocket payment, let's say for your copay if you go to visit the doctor. And there are some other additional measurements there as well. But in the PCE, it includes things like Medicaid, which is paid for through taxes by the government. That's out of the scope of the CPI.
So scope is really kind of the thing that differentiates these two indexes because one has a certain scope -- really consumers’ out-of-pocket payments, which is CPI. And the other has just a much broader scope and it's really able to capture more of what's happening in the economy and more of the inflation you see through the broader economy. And that's probably, I think, why you hear the Fed officials say that their preferences are for the PCE versus the CPI.
Tracy (06:04):
Can you talk a little bit more about that? So how did it come to be that the Fed is focused more on PCE? Going back in history, was there like an announcement or a trigger for them to focus on that measure versus something like CPI?
Omair (06:19):
Yeah, I don't know that there's necessarily any historical basis for it. I think it's just more that the PCE is more representative of the broader economy. You know, it just captures more of what the types of inflation that people tend to see across the economy. But it also captures some of the inflation that businesses are seeing as well across the economy.
Skanda (06:39):
If you don't mind me jumping in, there actually is a history to this…
Omair (06:42):
Oh great! Why don't you take that one?
Skanda (06:44):
So the Fed for the longest time actually did sort of focus more on CPI. They never had a formal inflation target until 2012. But if you asked the Fed how they're tracking inflationary pressures, they'd probably point to CPI first until the year 2000.
And so around that time, especially Greenspan, was focused on sort of the notion of quality change and how substitution bias might be at work, where the composition of what consumers consume changes, but CPI is conceptually more of a fixed basket relative to PCE, which is trying to dynamically change the weighting on the price index to match what people are consuming and put a little bit more emphasis on how consumer spending patterns change.
And so around 2000, the shift was from CPI to PCE within the Fed. The Fed didn't really say what kind of target was going to be placed on it. I think it was implicitly as deemed to be around 1% to 2%, 1% to 2.5%, sort of where Core inflation was.
But actually there was a very big difference in terms of will you choose CPI or PCE? There's obviously just an inherent bias in which CPI readings tend to be a little bit higher than PCE. And so that itself kind of changes sort of if you thought 2% with some magic number, it actually means different things if it's tracked in terms of CPI or PCE.
Joe (08:02):
First of all, that's really interesting and I didn't know that history. What's happening right now, just to get up to speed, because we're going to dive into the guts of some of these, but right now, there is a gap between PCE and CPI or the Cores. What are we seeing in the trajectory of this sort of, I guess you call it the wedge, the jaws between these two lines?
Skanda (08:26):
So typically the wedge between Core CPI and Core PCE to a first approximation, especially pre-pandemic, you would've said it was roughly 30 to 50 basis points in the year-over-year readings. So 0.3% to 0.5%, if you know what Core CPI is, you should be able to know what Core PCE is.
Joe (08:42):
With CPI being higher?
Skanda (08:44):
With Core CPI being higher, correct. And if you look right now, Core CPI year-over-year is something like 3.9% on a year-over-year basis. And Core PCE is going to track something around 2.9%, maybe 2.8%, somewhere around there.
Those readings are quite different, right? That's about a hundred basis point spread when the typical spread was 0.3 to 0.5%. And yeah, there's obviously a variety of factors that have led to that. I'll kind of let Omair jump in, if you were trying to sort of explain this -- the first major reason that sticks out to him.
Omair (09:20):
Sure. So, I think everyone normally focuses on the weights, right? And I think, Joe, you mentioned that earlier. And so they tend to focus on things like shelter inflation, right? And OER. And the weight of that in the PCE is only about 15%, 16%. But of course, in the CPI it's about 43%.
And so people tend to focus on that difference and say ‘Well, that tends to cause a big part of the wedge.’ And that's true in a very sort of static sense. And what I mean by that is that over a short-term horizon, let's say six months, seven months, eight months, those weights are just not going to change materially, right? That difference is just going to be roughly about the same over a short-term time period. And then the year-over-year rates obviously for OER don't move too dramatically either.
So whatever that wedge is coming from shelter inflation in one month will be roughly, you know, the same wedge in six months. What's been going on more recently, really the last six months, is that, you know, Skanda talked about the spread historically being 30 to 50. We had gotten down to about 45 bips in July. So we were very close to a historical norm between PCE and CPI.
In the last six months, it's blown back out to 1% and that's largely because the PCE has been slowing much, much faster than the CPI. And that's typically something, you know, people generally focus on the CPI as being the one that kind of causes the movements. But right now what we're seeing is actually, it’s the Core PCE that is just slowing much, much faster than the Core CPI. And that's blown the spread back out from about 45 bips or so in July to about one percentage point.
And there's more nuance, this isn't really about shelter, this is about some stuff happening in medical care. And really much more than that, it's really about the core services part of the story. And those are very, very different animals in the PCE versus the CPI. They're just not constructed the same in terms of the core services. And so that's really what we're seeing, is core services are slowing really fast in the PCE and in fact have kind of gone up a little bit in the CPI.
Skanda (11:28):
And just for your listeners to kind of get a sense, the CPI is source data. Which is to say there actually is a measure – they’re doing the direct measurement, the direct surveys of prices.
Better to think about PCE as a composite of different sources including CPI, but not limited to just CPI. We also learn a lot about PCE from PPI input data, not the PPI aggregates themselves, but specific inputs. As Omair kind of alluded to in healthcare, there are inputs there from PPI that matter for PCE.
And then there are things that exist outside of CPI and PPI that also matter. And it's that PPI and that other stuff that really feeds into Core PCE in a pretty meaningful sense and has driven more of this short run divergence even beyond what you might explain from kind of changes in weights, rent, Owners’ Equivalent Rent.
Tracy (12:35):
So I definitely want to dig into what's driving these various inflation numbers at the moment even more. But before we do, I have a sort of existential question, which is I'm always kind of amazed at how much mental energy we expend on the question of what prices are doing at the moment. And it seems like one of those things that like, alright, there's observable prices for everything. So why is it so difficult? And obviously, the weightings come into play here and what you choose to focus on, etc. etc. But does it matter if PCE is different to CPI? I mean, if we understand the difference in methodology, why should we care about this divergence?
Skanda (13:17):
I think it matters to the extent that you are trying to anticipate, let's say a CPI index if you're someone who's, let's say, is investing in inflation products and inflation-linked financial products. Or if you're somebody who's trying to focus on how the Fed is supposed to react when the Fed has kind of laid out an inflation target that's anchored to 2% on PCE and proxied by Core PCE over time, you'll get slightly different answers.
That's kind of the technical sort of market oriented answer. Actually, I think if you ask maybe some economists, they'll probably tell you ‘Well, CPI, PCE, PPI, what all does it really matter? It should average out. There is some underlying price level, these are all just imperfect approximations of the price level.’
But in practice, you get some pretty big divergences, especially in real time because of choices in weights, choices in methodology, choices in what scope of goods and services you're going to put more emphasis on or less emphasis on. And you'll get different answers.
And this is kind of how that like cottage industry of different private sector measures of inflation or Truflation or whatever it is, shadow stats, these [are] all various spectrums of crankery to less crank measures of inflation. But I think that's kind of the byproduct of the fact that all these choices do matter, right? And there is some level of faith to saying there's some underlying true inflation and that's actually here and it's not there. It just really depends on the choices you make.
Omair (14:44):
I would just add two other things quickly. One is just simply that I think we care about the CPI for two reasons also relative to the PCE. One is that simply put, it comes out first and it shapes expectations for where inflation's going. So, you know, we get that number typically by the 12th, 13th of the month, whereas PCE is the end of the month.
So I think for a couple of weeks people are digesting, you know, what is the Fed going to do based off of this inflation number? Even if we know their preference really is for the PCE, it just really starts to shape expectations very early on before we get the PCE.
And I think the second thing is that it's tied to real life in the sense that all the COLA adjustments made in social security, for example, are derived from the CPI. A lot of rent contracts are based on the CPI. So it does affect consumers in a variety of ways. And so I think that's why there's also still, you know, relative importance in terms of thinking about what the CPI is doing. In addition to what Skanda mentioned, which is that a lot of the PCE is built off of what the CPI is doing anyway.
Joe (15:43):
I want to get into -- and we're going to get into -- the sort of deep methodological questions and where these numbers actually come from, etc. But before we get into sort of complicated methodological questions, I want to ask a sort of simple methodological question, which is like, how does the government, say, in the CPI track the price of a tomato or something?
I don't know if tomato's a category, but something like that. Let's start with something simple because I get why measuring insurance is super complicated. And measuring various things like that [is] really difficult. But let's take something simple. Every month the government wants to know how much tomatoes and apples and pears cost. What is the basic process of collecting that information?
Omair (16:23):
Well, so there's the consumer expenditure survey conducted, well, now every year, it used to be every two years. Basically we're asking people to track their spending and what they're spending on and how much they're spending on these items. And there are thousands of these surveys conducted annually.
And so from those surveys, we're collecting data on what exactly, not just what kind of apple or whether or not they're purchasing apples, but what specific kinds of apples they're purchasing. And so based on all of this collected information, you are then figuring out exactly what you want to price and what the share of spending is for each item in the basket. So, you know, Fuji apples versus Granny Smith apples, whatever it may be. And these are done at, you know, the metro level. So the diary might be a little bit different, let's say in Chicago versus Miami versus New York in terms of some of these granular items.
So ultimately it's a consumer expenditure survey that's sort of dictating everything from what it is that you want to collect prices for, where you want to collect those prices as well, because you're also figuring out where it is that people are shopping, and then, you know, how much weight to put on an item within the basket. So apples within the food at home grocery store index, used cars within the transportation index and so on. So, that really dictates everything in terms of the CPI and how it's sort of being constructed and what's being picked to be priced.
Tracy (17:58):
Maybe this is a good point to talk a little bit about the deterioration in some of the survey responses that people have been discussing recently. So the idea that a lot of this is based on the proportion of survey respondents who actually get back to you and say ‘This is how much we're spending on Fuji apples per month,’ or whatever. And that percentage seems to be declining over time. I think I've written about this before, but I don't have the numbers in front of me. But it seems like that could be a pretty big deal for the accuracy of some of these figures or maybe just have some sort of influence on them.
Skanda (18:37):
Yeah, I mean I think it seems to be an issue because so many of these surveys are conducted over the phone, not everything is CPI is that way but…
Tracy (18:43):
No one answers their phone anymore. If someone calls me, I don't pick up.
Skanda (18:47):
I mean this is the same problem with election polling, right? Where we have a lot of non-response going up and you don't know if that skews a certain way. And if you're kind of calling particular businesses, there's a more structured process there relative to say election polling.
But there is still an issue, I think of, non-response that kind of permeates a lot of data. This is not just about inflation data now, it's obvious like labor market data has the same problem where response rates are going down. It's taking more calls to be able to kind of fill out the survey.
Relative to other data points, I'm not sure inflation data is actually as vulnerable, but this is a structural trend. The data is only as good as you measure it, right? So this is actually going to be an ongoing challenge for the BLS I suspect.
Joe (19:33):
So let's talk about some of these idiosyncrasies, particularly within Core PCE, because there are some things where you can just go to the grocery store, you can go to Whole Foods and you can go to Trader Joe's and you can go to Associated and Wegmans and look at the price of Fuji apples or Red Delicious -- God forbid, Red Delicious. Hopefully those are getting a smaller and smaller weight in the basket because they're the worst apple.
But then there are other things where you cannot do that. There is not just a price tag and things have to be imputed in some way. And so you have to sort of derive a price for things where there is just not a public price.
Let's talk about some of these imputations and where they get weird because I think this is where it gets like really interesting. You know, when it comes to, people pay financial services, they pay for a financial advisor, etc. I don't know [if] there's a simple price tag that that can be established. Talk about some of these more interesting or complicated or ethereal categories?
Skanda (20:33):
Yeah, there are a set of transactions, especially now as you get into PCE, where we have no real transactions we're observing, right? Probably the more common one would be sort of your Owners’ Equivalent Rent is very commonly cited, because that is effectively trying to approximate effectively the cost of rent for someone who owns their house.
And that is something that is not really a transaction there, but the BLS is, and the BEA, BEA obviously compiling Core PCE, they are basically using rent data to do that. So from the CPI housing survey comes the rent data. That rent data is also then used for estimating Owners’ Equivalent Rent. There's no transaction behind it, right? But it is basically saying ‘We're going to assume this price represents what Owners’ Equivalent Rent looks like.’
It gets trickier when you get into some other things in PCE, though, relative CPI, something like imputed financial services where there's no transaction taking place. It's specifically the value you, the consumer derived from your financial institution, your bank. When you are getting all these various services, free checking all sorts of other things, you're getting some value from whatever it is, Chase, Bank of America, I'm just giving examples. You're getting some services and yet you're also not being charged necessarily. You’re not getting sort of the deposit rate that reflects what like the bank itself earns…
Joe (21:52):
Well, you’re paying for it that spread between the bank deposit and Fed funds, right?
Skanda (21:54):
Precise, precisely. That that spread is a big part of the implicit price. And so the BEA is simultaneously trying to measure what's that volume of value that the consumer's deriving? And also the way of proxying,what's the price associated with that value? And these are two things that are basically dark arts, right?
It's kind of not something that follows an obvious and verifiable method for being able to say this is how much value you're deriving. There's just a lot of rough approximations , it is probably the funniest bit of Core PCE for my money because it is a function of two factors, which is the value you can think of as that spread between deposit rates and call it benchmark money market rates.
And so if the Fed is raising rates faster than deposit rates are moving, that's going to show up as more inflation. And we saw relatively high inflation from this category in 2022 and it really moved the needle on Core PCE even. And then on the other side of it, when SVB hit, you obviously started to see banks start to raise their deposit rates more aggressively and the Fed slowed down on hikes. And so kind of weirdly because of both of those factors, you've seen that part of PCE really ratchet down in the last six to nine months.
And this is all very weird stuff where it's basically a function o, well, Fed hikes are kind of inflationary through this category and then slowing down on Fed hikes and getting some deposit rate catch-up ends up being the opposite. It's kind of very bizarro stuff.
There's some other things that matter, but it just kind of goes to show you there's a lot of silly parts of Core PCE, I'd even say some parts of Core CPI. People tend to think the stuff is all very fundamental and mechanical and I just would caution that there's a lot of weird methodology and imputations that kind of go into various parts of that complex.
Tracy (23:42):
This is an interesting wrinkle to my campaign to improve the transmission of monetary policy by making everyone switch to higher interest bank accounts. But okay, I'm going to ask a slightly, or a very provocative question, but just on the question of how we measure shelter costs, can both of you choose, if you had to pick a preferred measure, would you be team CPI/BLs or would you be Team PCE and BEA.
Skanda (24:13):
Omair?
Omair (24:14):
I would probably say -- this is tough actually -- now that I think about
Skanda (24:22):
I mean on some level the BEA is using the BLS data. And so, the data itself about rent and Owners’ Equivalent Rent are coming from the BLS and the CPI housing survey and all of its perfections and imperfections.
I think the better question to ask is sort of like whether market rents are contracted rents. What we are basically measuring in CPI, what the BLS is measuring is contracted rents. So they tend to lag, they also tend to be a little bit smoother, a little bit more auto correlated, a little bit more obviously cyclical in a way that, let's say we had been using market rents this entire time, we’d probably have much lower inflation ratings now or we'd probably see something close to 2% on the Fed's key gauges and yet we would've had very, very high inflation observed in 2021.
Now maybe that's actually the true version of what was going on, but you also get, there's a bit of a trade off between getting smooth and cyclical versus something very volatile and jumpy and maybe not as reliable in terms of how to set monetary policy, if you think that operates with some lag.
So you would had either a super, super high inflation bulge in 2021 or something a little bit more smoothed out over 2021 to 2023, these debates kind of can cut either way. And I'm not sure which is actually necessarily better, but it's better to just appreciate the fact that there's a lag.
Omair (25:43):
Yeah, I'm going to come back and I'm going to say CPI and the
Tracy (25:47):
Very definitive!
Omair (25:49):
Yeah, the simple reason for that, I think in my mind at least, is the stuff that is quirky in the CPI, like we've talked about health insurance on the show before, the stuff that is quirky in the CPI, the PCE essentially just magnifies that even more by some of the items that are in that index, like the Financial Services Furnished Without Payment Index that, that Skanda talked about.
I think there's even more stuff in that index that is sort of, you know, conceptual and imputed. I mean roughly 13% of the entire Core PCE is just these imputed prices that no one sees. Which is why by the way, we also have a market-based Core PCE to get rid of that and just look at actual prices that people do pay and people do see.
So the CPI does of course have some of that, but I think there's less of that influence in the CPI than in the PCE. And yes, we can talk about OER being heavier weight, but at the end of the day that's in both indexes as well. So I would probably prefer the CPI over the PCE.
Joe (27:06):
Let's talk about more of these interesting categories. So one of the things that we saw, especially in the sort of, you know, the 2021, 2022, the sort of big shift in consumption patterns between goods and services, and we talked a lot about ‘Okay, goods are doing this and services are doing that and goods prices [are] through the roof etc.’
But one of the things that you've both pointed out over time, and I've heard you talk about this a fair amount, Skanda, is that this idea of drawing a bright line in many instances between a good and a service, particularly when it comes to measuring costs, is sort of impossible or fallacious in some way. And so auto insurance seems to be a big area in which we call it a service for measurement purposes, but it has connection to the goods aspect. Can you explain that a little bit further?
Skanda (27:56):
Sure. I think the conventional wisdom and the Fed has obviously done its job in trying to propagate this, which is to say that ‘Okay, the goods side of the economy, that’s supply chains, that's maybe a commodity shortage here or there. But services? Services sounds like labor, services sounds like wages.’
Service prices are obviously a function of what people who are working, how much they are paid and then it's just some spread on that. So wage growth should be dictating services is, it’s like an okay guide for maybe understanding even housing costs ironically…
Joe (28:23):
Right, like we think of services as like some a restaurant worker or a masseuse or someone painting your house or something that. And so it's very very linear to labor costs.
Skanda (28:34):
Yes. And so you think of someone who's maybe cutting your hair, right? That probably is more directly tied to one another, right? But at the same time, even like say housing, which I think of housing and rent as actually being related to the labor market, but it's not the cost of like construction or maintenance that's really dictating the cost of rent. It's the marginal supply and demand for housing. And that's something that's not really driven by labor in any sort of proximate direct sense.
But then we get to a lot of things like vehicle insurance and we get to airfares. Those are areas where you can definitely point to specific things where goods and good supply chains really matter.
So jet fuel costs are very important because they're the most variable cost in an airline’s cost structure. Quite reliably, jet fuel costs feed into what airfares look like. So when jet fuel spiked in 2022 due to sort of the invasion of Ukraine, you saw obviously crack spreads blew out, oil prices blew out and therefore jet fuel prices blew out. And that had a pretty reliable and predictable pass through into airfares. It even showed up in previous instances, like the 2008 oil price spike having a pass-through effect into airfares.
In the case of vehicle insurance, right, and it's not just vehicle insurance -- we can think about motor vehicle leasing, repair, maintenance, rental, these are all what I'd call value sensitive services. So the value of an automobile will shape the cost of being, that also affects the price of parts, right? Because let's say automobiles are in shortage, then the value of repair goes up because of it. The value of spare parts goes up.
And we've basically seen that and that's obviously affected the cost structure for insurers. There are some other regulatory dynamics going on with respect to vehicle insurance and that also has an implication for CPI that's different from PCE, but the connection between goods to services is itself like kind of important to appreciate.
Yes, services tends to move more slowly with more of a lag, but it is kind of fundamentally moving with respect to a lot of these dislocated supply chain and commodity supply issues in ways that I think are in some ways just removed from the labor market itself. I think that is something the Fed has been kind of keen to say, well, the labor market must matter somewhere, so let me jam it into all these other services aside from housing.
In practice though, there's just a lot of stuff even in supply chains and commodity price swings that actually have a lot of relevance beyond what we strictly classify as a good and has a lot of impact on the services side of the economy, and services that are consumed but maybe [are] actually pretty capital intensive or not necessarily tied to the direct price of labor.
Tracy (31:10):
That reminds me actually, can you talk a little bit about the timing of some of these pricing decisions? And what I mean by that is I remember talking to Omair about producer prices for mayonnaise, I guess this would've been over two years ago. And there was an idea there that companies often revise their prices around quarter ends. There's also a more recent phenomenon that I think the Goldman Sachs analysts are calling the January effect, this idea that, well, in a new year lots of people revisit how much they're charging and unveil all their new prices around the new year. But talk to us about the actual timing and mechanics of how prices get changed.
Omair (31:54):
So I will say that there is, so this so-called January effect, which the rest of us just call residual seasonality in the data, that's been something that's been prevalent in the start of the year. Price increase is something that's been prevalent in the CPI data really for probably the last 20 years or so.
And you tend to see it much, much more in goods. We're talking more about furniture, apparel, things of that nature where contracts for delivery tend to get reset for the start of the year. And so what you'll typically find is that, at the very start of January and February, you tend to see prices on an unadjusted basis increase by more than what you'll see them do over the balance of the rest of the year.
And that's something that's been pretty well known in the CPI data and the idea is the seasonal adjustments should be able to sort of offset that seasonal move every year. But of course the problem is that, you know, those price increases aren't sort of a static thing. They tend to move around quite a lot. And so seasonals never quite are able to capture it in that first go. It's only sort of several years later when the seasonals are kind of redone on those particular years where that that effect kind of compresses in January.
But we saw it, you know, this year for sure where we had some big moves. What was really interesting though is, so typically residual seasonality, the way it works is that first quarter, the Q1 data is typically stronger than the rest of the year. Two thirds of that strength is in core goods and only about a third of it is coming from the services categories, which makes a bit of sense when you think about, you know, delivery cost changing for shipping goods items across the country and from overseas. That gets worked into new contracts at the start of the year for goods.
What was really peculiar this year though is that goods really didn't do much, you know, core goods even excluding autos, all the strength was really in the services categories and that's what really kind of stood out this January was it wasn't a traditional, you know, it wasn't your father's residual seasonality. This was something a little bit different because it was so heavily focused on the services side.
And so that's something I think we just need to be careful of going forward is normally when you think about residual seasonality, you say ‘Okay, it's a January February effect, once we get into the second quarter it'll go away.’ And if it was in core goods again this January, I would've said ‘Yes, that's probably the right take.’ But I would just say there's maybe a little bit more caution here needed because it wasn't in core goods, it was in core services. So I think that's just something we need to be a little bit careful of thinking about the data going forward
Skanda (34:27):
And something that may be an extension of Omair’s point here, so Omair’s talking about core services and CPI, this is exactly where thinking about the wedge is especially important because when you think about the particular prices that are of relevance for PCE, once you get outside of goods and housing, goods and call it rent and Owners’ Equivalent Rent specifically, there's a pretty big divergence between what you learn from CPI, especially in that kind of Super Core, core non-housing services.
CPI is very different from core non-housing services PCE, and they're just like, we talked about imputed financial services as being one goofy example, but there are other examples including vehicle insurance and airfares where they're just measured in different ways. And so the residual seasonality that shows up in one part of a core service CPI segment doesn't necessarily have a neat cognate core services…
Joe (35:23):
Wait, how are they measured differently, those categories in the two in indices?
Skanda (35:28):
Which categories?
Joe (35:29):
Well, I think you said airfares and motor vehicle insurance are measured differently?
Skanda (35:34):
So motor vehicle insurance for example, is more directly tied, insurance products in general and the CPI is really about a function of payment and what's really the out-of-pocket cost to the consumer conceptually. So it's really tracking that in real time and obviously right now you have a lot of, state by state insurance companies are pushing for regulators to allow them to charge higher premiums for auto insurance after a sort of a cycle of I think a lack of profitability relative to the cost and expenses for insurer.
Whereas for vehicle and for insurance products and PCE, it's really about the value that is trying to proxy or capture, right? So the value to the consumer is a function of not just what you pay, but also what third parties are paying and what you're getting back in terms of the expenses that have to be covered.
So they do tell you different things there. So vehicle insurance and PCE has been much more benign. It comes from a PPI segment, whereas vehicle insurance in CPI has been very strong. And so that's like one part of the divergence. Then there's another one: airfares, is actually a function of, I think Omair can really speak to this with high expertise here, but the CPI is tracking very specific routes and trying to track them systematically over time in terms of how much it costs to be able to fly from one place to another, different categories of seats on airplanes.
But in PPI the, it's meant to reflect sort of the revenue per passenger mile and it tends to have a little more of an upward bias actually, relative to CPI. So PPI for airfares has generally run stronger than CPI and especially so, so despite the wedge blowing out, it hasn't been a function of airfares as much as other categories, but I'll let Omair expand further
Omair (37:13):
Well, I was just going to actually just quickly put some numbers on the auto insurance thing, because yes, the methodologies are different, but to give you some sense of it, auto insurance in the CPI right now, year-over-year, is running at about 21%. In the PCE is running at about 8.7%.
So methodologies are different. They produce vastly different numbers and growth rates for these two indexes. So given their weightings, you're going to have just a very different influence of what is ostensibly it seems like on the surface the same thing -- auto insurance -- but the different measurement is resulting in just vastly different growth rates, which means its impact on the core and core services overall is also just going to be hugely different. So that hopefully kind of gives some context around what the difference in methodology can mean for the growth rates around these particular items.
On airfares, yeah, you know, Skanda mentioned, the CPI is basically very much just a weighted average of routes, you know, Chicago to LA, LA to Vegas, what have you, and is very much just capturing directly from the carrier's website what it is that you're paying for that flight. And also the cost of the first check checkback.
And the PPI is also looking at these routes, but looking at the overall revenue per passenger mile as Skanda mentioned. So month to month they can actually differ quite a bit over the longer term. The trends, directionally they tend to be in the same direction, but it can cause , variability on a month-to-month basis where you could have the CPI, for example, up three 4% and the PPI actually down a couple of percentage points. So on a month-to-month basis, you do have to be sort of be aware of what these differences are because they can produce pretty different results and pretty different impacts on the relative cores.
Skanda (39:02):
I mean airfare is a really big one in terms of CPI, just because it's so volatile in the CPI and you'll typically hear right afterwards, well, core services ex-housing CPI was really strong or really soft and it's driven by airfares. And that just doesn't have the sort of bearing on that sort of Super Core PCE component because that's coming from BPI. And you can learn that usually only a day or two later.
Omair (39:25):
I can't tell you over the years the amount of time and money I've spent trying to get airfares right. And it's worth like less than 1% of the Core CPI because, again, the, the volatility, it could be up 8%, down 8% and so it can add or subtract, you know, five to 10 basis points each month from the core.
And so getting that, it's just one of these high volatility ones where, yes, it's not worth the lot, but the magnitude of the moves is so great that it just demands a lot of attention. Hotel rates as well, same thing.
Tracy (39:54):
Gosh darn airlines and their dynamic pricing! So we mentioned earlier that we're recording this episode right before PCE comes out. Can I put you both on the spot and ask you for PCE guesses?
Omair (40:06):
I am probably, I think a little bit below some of those [forecasts] -- so most of, I guess the forecasters I would want to follow are right around 0.40. I'm a touch lower at 0.36. I think the difference is just assumptions about imputed prices that people are making.
But on the Super Core PCE, I'm expecting a 0.5. So that is going to be a pretty strong print. I think obviously on the Core CPI we had about a 0.85. But yeah, I think it's going to be a relatively firm print. The PPI data was strong, the CPI data was strong, so at least for January it looks we're going to get a pretty robust move in in the PCE data.
Skanda (40:45):
Yeah, this is actually coincidental because I'm not sharing my work with Omair on these things, but we ended up in the same place on both Core PCE being 0.36% and Super Core for myself, 0.5%.
But I also will note, for at least on the year-over-year reading, something to appreciate is there's also revisions. So while some people might have higher month-over-month readings, they may not also reflect the fact that there are reasons for revisions to prior months. So for example, and it's something I think also Omair will be here to talk about too, which is financial services prices now, not the imputed ones, but the ones that are more measurable, those prices also kind of factor into revisions and those are why I suspect that we're probably going to see more shallow inflation progress. So if we kind of were around high 2.9%, low 3% last month, that's actually going to bump up a bit. And at the same time we're going to see 2.88% for Core PCE…
Joe (41:38):
Real quickly, I'm glad you said, this is going to be my last question. Explain measured financial services. So when stocks go up, measures of inflation go up?
Skanda (41:46):
Correct. To a first approximation, measured financial services are a lot of the portfolio management services that consumers are effectively consuming. And those, the way they're measured tend to, in the aggregate, kind of proxy what the equity market's doing. How much so can vary a little bit over time, but the equity market does matter. So the equity market's been up a bunch over the last few months and that has taken time to really show up in the PPI data. But it has, and that PPI data kind of is relevant for sort of PCE purposes.
Omair (42:17):
Yeah, and in the PCE, that portfolio management index, it added quite a bit to the January Core PCE number. I think it's going to add something like eight basis points to that number alone. And that is mostly a reflection of how the equity market did in Q4. So up around I think S&P was something like 12%, 13% higher. That pretty much goes into the returns that are reported to the BLS.
So these equity returns from mutual funds and ETFs and private portfolio managers, they come back and say ‘Hey, we did great in Q4, here are our numbers.’ That gets built into the PPI data, which then flows into the PCE.
And so the index is worth about one point a half percent, but when you're up 5% or so roughly as it was last month, you're going to get a big pop in the Core PCE. So that's likely what we already saw it in the PPI and that's what's going to feed into the PCE. So equities do matter quite a bit for that number.
Joe (43:15):
All right. Well, this episode comes out at 4:00 AM Eastern on the 29th, so people have four and a half hours to listen to it and then, you know, understand what's going to come out at 8:30 AM Eastern.
Omair and Skanda, thank you so much for coming on. That was fascinating. We want to do more of these episodes where we actually learn about the data that we talk about all the time. But appreciate you both for coming back on Odd Lots.
Omair (43:38):
Thank you. Thanks for having us.
Joe (43:53):
Tracy, I found that to be a really interesting conversation. I have to say, you know, the cranks will hate me for this. I have a lot of admiration for the public officials, the bureaucrats, the economists at the BLS and BEA, who have to assemble all this stuff and figure out , you know, like the different ways of measuring airline costs and portfolio management costs. It doesn't seem easy.
Tracy (44:17):
I think that's a totally fair thing to say. The other thing I would say is, the people doing this, at least at the BLs, are really responsive to inquiries. And Omair would be able to talk about this and maybe we should have asked him, but if you send them a question, and ask -- I did this for that mayonnaise story and asked them ‘How are you calculating this particular line item?’ They will get back to you. And they'll get on the phone with you and explain it for 20 minutes.
Joe (44:44):
This is true. And actually listeners should know this…
Tracy (44:48):
Wait, don't flood the BLS with calls just to test this theory...
Joe (44:50):
No, don’t. But when you see someone and you see something, they're like ‘Oh, look, this is crazy. Look what they hid in this area.’ They're very transparent and this is true. You can call them up and they'll say ‘This is what we saw.’ This is how it works, so Tracy's making an excellent point. No one's going to take us up on that because people prefer to believe conspiracy theories. But if you see something that seemed [strange], they will respond.
Tracy (45:14):
If you see something, say something in inflation data?
Joe (45:16):
If you see something, if you see a weird number, call them up and they will explain how it came there.
Tracy (45:21):
Well, the other thing that comes out of that conversation, and again, I thought both Skanda and Omair were incredibly clear in laying out the difference in methodology, but really the overarching theme is the amount of subjectivity and value decisions being made when it comes to how to present this overall data.
And you can get super granular on this and think about all these things that are sort of hidden in the background. But I remember there's stuff like, you know, the weighting of certain items is different depending on where you are in the country. There's also qualitative adjustments. So your refrigerator, maybe the price is declining, but now it comes with WiFi and so you have to incorporate that qualitative judgment as well.
Joe (46:04):
Totally. Well, and some of these things, and Omair made this point about, I think in one of the insurance categories about how CPI and PCE are different. So you could imagine, let's say someone pays, I don't know, a hundred dollars a month for car insurance and then the next month, the next year it goes up to $110. So that's a 10% increase. But let's say that in that second year, you're getting $107 back every month equivalently in repairs for your car and the previous year you're only getting $90.
Well, maybe arguably your insurance just got cheaper because the amount you're getting back relative to the amount you're paying is so much higher. And so you could see immediately how something that's very important – insurance -- huge part of the economy, becomes incredibly difficult to measure when you're trying to gauge well, are you just trying to gauge how much you pay? Or are you trying to gauge how much you paid relative to the value of the service you received, which is going to change based on the value of your car and the difficulty of obtaining the replacement components. It's really tricky stuff.
Tracy (47:08):
Well, I also liked Skanda’s point on this note about deposit betas and the impact on inflation...
Joe (47:13):
Oh yeah, I love that. Inflation's so wild. That blows my mind.
Tracy (47:15):
I know, I feel we should end this before we start talking about how gas prices going down is actually inflationary because people pay more…
Joe (47:22):
Oh, I thought you were going to say we should end this before I suggest that we should be paying our banks for checking services.
Tracy (47:26):
Well, that too.
Joe (47:27):
But no, it really clicked. Because, you know, we had that episode with Steven Kelly and I was like ‘Well, what is the service of a bank? And he's like ‘The service of a bank is deposits and that's why you get that cheap [deposit] rate because they're providing you a service.’ So there is this sort of intellectual logic behind, okay, you're getting 2% rates for your checking account Fed funds [is]at 5%, therefore implicitly you're paying 3% for those checking services and all those other things. But these are tricky things to measure. But it is funny how mechanically, just like the Fed raising rates and increasing that gap between rates and deposit rates just sort of mechanically increases measured inflation.
Tracy (48:06):
Joe, you're about one connection away from breaking out the yield bug hat again.
Joe (48:12):
I'm like that, what’s that meme of the guy? I think it's Always Sunny In Philadelphia. That's where I am right now.
Tracy (48:19):
Okay. Shall we leave it there?
Joe (48:20):
Let's leave it there.
You can follow Omair Sharif at
@fcastofthemonth
and Skanda Amarnath at
@IrvingSwisher
.