How the US Measures Inflation
Introduction
Inflation is an elusive killer. It silently lurks in the background of economies and is, to many, only measured subconsciously by the pain their bank accounts slowly endure over time. And while pain is perhaps the most real measurement anyone could ask for, it helps to attach a more tangible number to this money murderer.
In a vacuum, measuring inflation is simple: How much did that thing cost, and how much does it cost now? There’s your inflation. That approach is obviously still valid, but there are hundreds of thousands of goods and services now, all of which have their own price moves. In addition to all those different rates of inflation, there are millions of people with completely different spending habits, so it isn’t always clear how to decide on a specific number or choose a basket of goods that applies to everyone.
This is because inflation isn’t the same for everyone. Everything you buy from groceries to clothing to gas to amazon orders to everything else creates a very specific inflation rate only applicable to you. Depending on your specific basket of goods, your inflation rate could be 1% or it could be 12%.
When anyone measures inflation, then, they’re either measuring their own rate or they’re measuring some kind of aggregate inflation loosely applicable to everyone. This is not to say an aggregate inflation number doesn’t paint a decent picture for all or most consumers, but it should always be taken with a grain of salt because it’s impossible for an aggregate inflation number to be a completely accurate representation of your individual inflation rate.
That said, the institutions charged with generating aggregate inflation numbers generally make a valiant effort to find and measure goods that apply to most people. Pretty much everyone lives in a home or apartment and I’ve been told that everyone eats food, so it’s not as if there aren’t some obvious inflation rates to track.
There will naturally be a lot of disputes over the specific methodology for calculating inflation, and rightfully so, but the gold standard to always keep your eye on will be the inflation numbers the federal government issues.
Whether the Fed’s methodology for measuring inflation is the most accurate or not, the inflation number the government determines is the most important. Not only is it the number the government itself uses to inform its policy decisions, but it’s usually the number the economy and markets reference in their own calculations, so it is by far the most relevant.
Knowing just how important the Fed’s inflation numbers are, it should be of similar importance to know how exactly the Fed is coming up with them. Understanding how the Fed arrives at its numbers will add a great deal of context to the headline number we all see, and it can also help to connect various dots within the overall inflation narrative, which could offer an edge in trading the market.
With that in mind, let’s walk through how the federal government actually measures inflation.
Consumer Price Index (CPI)
In the United States, the Bureau of Labor Statistics (BLS) issues the most widely used measure of inflation. Through two surveys of American families, one that records prices of most goods and services (about 80,000 items), and another recording the price of housing and shelter for about 50,000 residents, the BLS creates the Consumer Price Index or CPI.
Through these surveys, families will help the BLS determine which items should be added to the inflation basket and how much weight to each of those items it should assign. The BLS will try to weight each item according to how important it is to the average consumer they survey. As an example, if consumers are spending more on gas than clothes, the BLS will make it so the changes in the price of gas have a greater impact on the CPI than the changes in the price of clothes. That said, the BLS only updates the CPI spending weights every two years, so there can be a significant lag in tracking shifts in consumer spending.
As for the housing and shelter portion of the CPI, the BLS takes those roughly 50,000 Americans surveyed and breaks them down by renters and owners. If someone is a renter, the BLS simply counts the rent paid to the landlord and any utilities included in the lease. However, if someone is an owner, the BLS will use what they believe it would cost them to rent out their home rather than measure the actual price change in the value of the home, which is known as Owner’s Equivalent Rent or OER.
The BLS will ask the homeowners it surveys “If someone were to rent your home today, how much do you think it would rent for monthly, unfurnished and without utilities?” As you can imagine, answers to this question are bit more subjective than scientific, which is not particularly reassuring once you realize that OER alone accounts for about 24% of CPI.
The reason the BLS records home inflation this way is that they aren’t in the business of tracking asset appreciation, which is what a home value increase is. The BLS only wants to track consumption. While there may be a sliver of merit to the OER approach, it is definitely a convenient distinction to avoid massive surges in CPI due to the even more massive increases in home prices. If the BLS didn’t use OER as its input, CPI and home inflation specifically would be recorded at much higher levels.
Chained CPI
In addition to the traditional CPI, the BLS has created a variation known as the chained or chain-weighted CPI. The main difference between the two is the chained CPI uses more current data for consumer habits and it measures how consumers respond to changes in relative prices. The idea is that consumers will adjust their spending habits to purchase more of a similar good whose relative price has declined against another good they used to buy more of.
For example, if consumers used to buy a lot of steak but steak prices inflated too much, they might shift to buying hamburger meat. Both goods are essentially interchangeable as far as the CPI is concerned, but one price has inflated less, so consumers buy more of it and the chained CPI changes its weighting. This is known as the substitution effect in inflation measures and is one of the ways the government can try to hide inflation in certain markets.
Just because hamburger meat hasn’t inflated as much as steaks and consumers were all but forced to switch their spending habits doesn’t mean the steak inflation isn’t real or shouldn’t be counted. The shift in spending would be a direct consequence of the inflation in steak, so disregarding it by changing the steak’s weighting in the CPI basket as consumers adapt to the inflation seems unfair. However, that is essentially what the chained CPI will do by swapping the goods it measures in each respective category.
Chained CPI might track specific consumer spending habits a little bit better than the regular CPI, but it’s hard to fully trust it when higher inflation can actually lower the headline number just because consumers switch to buying something else they can actually afford, but I digress.
With a basic framework for how both headline and chained CPI work, let’s dive into the basket weightings to see how goods and services are broken down.
CPI Category Weight Breakdown
Although the CPI category weightings change every couple of years as new consumer data comes in, habits generally don’t have dramatic changes, so the most recent weighting breakdown will still offer a good idea of how the basket of goods normally looks.
The largest category in the basket is shelter at a little over 30%, followed by food at about 14%, transportation commodities (including gas) at about 8%, energy also at about 8%, medical services at around 7%, and a little over a dozen more categories from 6% to less than half a percent.
Because shelter and food are the biggest categories, price changes in those goods will affect the inflation numbers a lot more than price changes in something like apparel, which only sits at around 3% of the total basket.
That means shelter inflation, such as increasing rents or OER, will significantly overshadow other inflation or deflation in the basket. It also means the headline inflation number may not always paint an accurate picture of inflation in all areas of the economy if and when shelter prices have large price moves. That said, shelter is usually people’s biggest expense, so shelter inflation will have the biggest impact on consumers.
But enough about CPI. Let’s transition into some of the other important inflation numbers and baskets the government uses.
Personal Consumption Expenditures (PCE)
Although CPI gets the most press and is referenced more for rent, social security payments, or other economic adjustments, the PCE is what the Fed refers to when talking about its target rate of 2% inflation each year.
While CPI and PCE are very similar, there are a few key differences to keep in mind. One is CPI is determined by the BLS and PCE is determined by the Bureau of Economic Analysis, but who really cares about which boring government institution is responsible for the numbers.
One of the main differences to focus on is the CPI is based on a survey of what all urban households are buying and the PCE comes from surveys of businesses. That means the data they collect is based on actual spending and sales rather than consumers who just relay what they’re buying through surveys.
The expenditure weights in the PCE are also much more flexible by changing as people substitute goods and services for others. While chained CPI employs something similar, headline CPI is much less fluid in response to changing consumer preferences. CPI also includes less comprehensive coverage of goods and services because the PCE takes its data directly from businesses and corporations that have a deeper insight into the full scope of products that are sold.
The last key difference is the CPI only covers out-of-pocket expenditures on goods and services and doesn’t include expenditures like medical care that aren’t paid for directly, but those expenses are accounted for in the PCE. This means expenses your employer or anyone else pays on your behalf are counted in your expenses in the PCE.
Generally speaking, PCE numbers are lower than CPI numbers because CPI doesn’t adjust as much to changing consumer habits (the substitution effect). The PCE, on the other hand, tries to track what is actually purchased and represents how consumers change their buying patterns when relative prices change. As a result, the PCE will typically have smoother price changes and lower levels of reported inflation. That said, some people believe the PCE is too broad and still prefer the CPI because it provides data that is directly reported by consumers.
With anything this complicated, broad, and inherently subjective, I would advise you to simply pick your poison. That doesn’t mean ignoring one number altogether, but if one methodology resonates with you more, so be it.
Before you make that decision, however, let’s take a look at how the PCE weightings differ from the CPI.
PCE Category Weight Breakdown
Just like CPI, PCE has its own weightings that influence its headline number, but they are broken down in notably different ways. PCE is divided into services, which is about 64% of the total basket, and then goods for the remaining 36%.
Within the service category, housing sits at about 16%, healthcare at 17%, hotels and restaurants at 6%, transportation at 3%, and so on. As you can see, this is already very different from CPI, especially with regard to housing and medical care, which is around 30% and 7% of the basket in CPI, respectively, compared to 16% and 17% in PCE.
As for the goods category, PCE has food at about 7%, auto at 4%, energy and gasoline at 3%, clothing at 3%, and so on. Again, very different than CPI, which has food at 14% and energy/gas at 8%. It’s hard to say which of the two has a more accurate weighting for any good or service, but these differences give you an insight into just how different different inflation numbers can be.
And as if all this wasn’t already confusing enough, there are also subcategories of these inflation measures called core inflation that exclude certain baskets of goods.
Core Inflation
Core inflation is a distinction between headline inflation and applies to both CPI and PCE. While headline numbers include aggregate prices for all goods and services, core inflation strips out the more volatile food and energy prices.
Over the short term, core inflation may give a more accurate reading of where inflation is headed, but people do buy food, fill up their gas tanks, and heat their homes, so headline inflation will more accurately represent people’s actual expenses, which might seem obvious when you realize how important food and energy are in everyone’s daily lives.
Because core inflation cuts out food and energy prices, it usually comes in a little lower than headline inflation numbers. That could make it sound like a cheap way of lowering the inflation rate, and that might be partially true, but the purpose is to cut out volatility so the government can make more accurate predictions to inform policy.
Even though food and energy are indispensable goods, their tendency toward large price swings can make it challenging to accurately forecast, so core inflation is just another number economists can refer to in their calculations.
With the various inflation baskets and basic weighting methodologies now outlined, let’s get into some of the specific adjustments beyond core inflation that the government makes to inflation data and the headline numbers we ultimately see.
Seasonal Adjustments
The first adjustment I wanted to discuss is seasonal adjustments, which remove the effects of recurring seasonal influences throughout the economy. To be more specific, seasonal adjustments will try to account for price changes related to things like changing climate conditions, production cycles, holiday spending, business sales, or the like.
As an example, a good like an orange is available year round, but prices are significantly higher in the summer months when suppliers are between harvests. Another example might be running shoes increasing in price during the summer when more people go running. The goal of these adjustments is simply to eliminate the effect price changes that occur at the same time and in about the same magnitude every year have on inflation numbers.
Seasonal adjustments might seem a bit fishy on their face, but if the costs of certain goods and services predictably inflate or deflate every year because of holiday shoppers, natural crop cycles, or the actual weather changes throughout the year, it makes sense to account for that.
The magnitude and regularity of the changes are the key metrics to keep in mind, but as long as those are consistent, there is a strong argument to exclude the effects those changes have on prices in certain months or times of the year.
Now let’s talk about one of the other adjustments the government makes to inflation numbers: hedonic adjustments.
Hedonic Adjustments
One tricky aspect of inflation is that products and services don’t maintain a constant level of quality, meaning their price fluctuations may not be solely due to inflation.
The most common example of this is technology. Although the price of an iPhone might go up every year, we are usually getting more features or value. A 10% increase in price could easily be offset by a 10% increase in the product’s overall value, so this idea of hedonic adjustments was added to inflation calculations in the year 2000 to help account for these potential discrepancies.
Determining what amount of value was added to a product that might offset its price inflation is obviously very subjective, which poses its own problems, but the real issue is these adjustments don’t always work the other way around. Even if you believe the government will fairly and accurately calculate hedonic adjustments, its inflation numbers rarely take into account a reduction in quality.
I’m sure everyone has, at some point, experienced a product or service they regularly buy slowly get worse over time despite its price staying relatively the same— or maybe even increasing. After all, if improvements in products are possible, surely deterioration is equally possible and present in the economy. That seems like common sense and yet the government tends to focus the vast majority of its adjustments on increases in quality.
Whether you agree or disagree with the inclusion of hedonic adjustments in the inflation numbers, and I would argue they’re completely fair to include, I think everyone would acknowledge that including adjustments in the opposite direction is pretty important if you do decide to take the hedonic route.
This is not to say that the Fed’s inflation numbers are completely wrong or that these adjustments drastically overstate the quality improvements in the economy, but it’s important to be mindful of the self-serving ways the Fed can bend or shape inflation data.
And this brings us to a very relevant distinction: How did the government use to measure inflation?
How the Government Use to Measure Inflation Data
Fixating on the old way of doing things might seem irrelevant, but understanding the different methodologies that return different numbers is vital for adding context to the inflation debate. Inflation is frequently viewed through the lens of history, but an accurate lens requires a consistent unit of measurement, which we don’t have because of all the changes made over the years to our inflation measures.
On a more granular level, consumer spending habits and, as a consequence, the basket weightings, were undoubtedly very different in the past. That alone isn’t a reason to do away with the comparisons, but it’s at least a small asterisk that should be noted in any comparison. In other words, we should probably take a deeper look at the specific inflation rates of goods rather than focus on headline numbers given how different the basket in any given time period could be.
But enough with the small potatoes critique, let’s get into the two major differences that warrant serious questioning when drawing comparisons between current and past inflation numbers.
For starters, hedonic adjustments weren’t introduced until 1998, so comparisons between present day and any time prior to that should, at the very least, ignore the influence of hedonic adjustments on current numbers.
The second problem, however, is even more of a concern and that is OER. The OER approach to measuring housing inflation wasn’t introduced until 1983, which was right after the crazy inflation of the late 70s. And given just how much OER affects headline numbers, failing to account for that difference when comparing today’s numbers to the inflation numbers from the 70s is tantamount to absurd.
If you have a dramatic change in how the highest weighted good in the inflation basket is measured, that is going to have an outsized effect on the headline number, meaning we should be talking about it if any historic comparisons are to be drawn.
All that being said, this is by no means a criticism of the revisions made to the CPI or PCE over the years. Most of the changes are rather sensible, but to ignore those changes is to ignore a crucial detail when comparing current rates to the past.
Everyone today loves to talk about the 70s and how inflation was so much worse back then, but they are clearly comparing numbers that were not measured in the same way. That would be like measuring something in centimeters one year and inches the next and then reassuring people that the inches number is lower than the centimeters number so we have nothing to worry about.
Again, this is not a statement about the merit of the changes the government has made to how it measures inflation, but it or anyone else using its historic data should at least try to apply those same changes to previous data if they are going to make apples-to-apples comparisons.
It’s also important to note that any discrepancy between current CPI and CPI of the 70s doesn’t necessarily mean inflation today would come in at 15% like in the 70s. In fact, it could just mean that inflation in the past was incorrectly measured and should have been much closer to current levels.
Now, if I was backed into a corner, I’d probably say the real rate of inflation in both periods is somewhere in the middle, but that doesn’t really matter because the point is they weren’t measured using the same methodology and so they shouldn’t be talked about or referenced as if they were.
The level of gaslighting that has been going on in recent months about inflation when it comes to these comparisons is unreal, so people just need to keep in mind that current levels of inflation cannot be fairly compared to the 70s or most other historic rates without accounting for the various changes.
How the Market Measures Inflation
Alright, I’ve dug into how the Fed measures inflation, but there is also a way the market will measure inflation that you can constantly refer to throughout the day. These measurements are broken down by time periods like 10-year inflation expectations, 5-year inflation expectations, etc. However, they won’t give you anything below five years, including spot or month-over-month inflation.
If you want short-term inflation numbers or expectations, you’ll have to rely on the Fed, do your own research, or take a deep dive into the yield curve. That said, the inflation expectations the market does provide are extremely important and something the Fed strongly considers when making monetary policy adjustments, so they are still key metrics to watch.
The most popular measure of the market’s inflation expectations is the difference between the yields on nominal Treasury securities and Treasury Inflation-Protected Securities (TIPS). If, for example, the US 10-year bond was trading with a yield of 2.5% and the US 10-year TIPS was trading with a yield of -1%, the market would be telling us it expects inflation to average 3.5% over the next ten years.
This difference between the nominal Treasury yield and its TIPS counterpart would also work for calculating the 30-year or the 5-year inflation expectations. And if you don’t feel like doing math every time you want to measure inflation expectations, you can look up the inflation breakevens for each duration that will calculate each tenor’s respective difference for you.
As an investor, inflation breakevens are critical because they offer a fluid measure of inflation that adapts to market expectations on a daily basis, so I would highly recommend adding them to one of your watchlists.
Conclusion
So I realize that was a whole lot of words and numbers, but what should you take away from all this information about inflation? Hopefully, at the very least, a deeper understanding of how it’s calculated in its many forms, but also that no single inflation number is *the* inflation number.
Calculating inflation is complicated when your aim is to represent the inflation rate of hundreds of millions of people, so no institution—government or otherwise,—is going to get it just right. And that isn’t necessarily a knock on the government or anyone else for trying, it’s just an impossible task.
Although there are certainly some weak points and shady tactics when it comes to how the government calculates CPI or PCE, both still have a thoughtful approach that delivers a reasonable aggregate inflation rate. Any number the Fed issues will almost certainly understate the real average inflation rate to some degree, but it will still tell you which basket of goods inflation is hitting the hardest and also give you a general idea of its rate of change.
Even if CPI or PCE numbers are vastly understating the real rate of inflation, they will understate it in roughly proportionate ways throughout each basket, so you can still trust it’s telling you what areas of the economy to watch. Knowing where inflation is can sometimes be more helpful than knowing the actual rate, so that first point is far from trivial.
The second point about the rate inflation is changing is also extremely relevant because the rate of change—or the second derivative as it’s called—is what the Fed is often most concerned with.
When it comes to implementing monetary policy, an increasing or decreasing rate of change in inflation demonstrates to the government that its policy and forward guidance are either working or inadequate. And whether you believe the Fed actually has an effect on inflation or not, anything that motivates the Fed should be of principal significance to investors.
Bearing all this in mind, there are some key distinctions that need to be made in anyone’s quest to unravel inflation. If you are an investor, you don’t need to accept the headline numbers to let them inform your decisions. The Fed and markets react to the numbers the Fed issues and they paint a good enough picture of the direction, speed, and location of inflation in the economy.
However, if you are solely focused on an accurate headline number, you’ll probably be chasing a lost cause. If the government can’t give an accurate headline number with all its resources and institutions, you probably can’t either, but that doesn’t mean you can’t calculate your own inflation rate.
Each person’s individual rate is what really matters. Your inflation rate may not be applicable to others in the economy, but it will give you a better idea of how good or bad the economy specifically is for you.
I would recommend sifting through some of the Fed’s numbers and comparing them to the inflation you personally see in your life to get a small insight into the accuracy of its headline number as well as the accuracy of its individual baskets like food or housing.
Your personal inflation rate, as a sample size of one, is way too small to consider actionable from a trading perspective, but it can open your eyes to potential trends the government’s numbers are too slow to capture, which is a great boots-on-the-ground way of getting a leg up on other market participants.
However, in the end, the most important thing to remember about inflation is that there is no single rate. There are trends, fundamental drivers, and similarities among all inflation rates, but inflation is ultimately measured by every individual person and their unique expenses.