Related Links

• An Alternative Measure of Core Inflation: The Trimmed Persistence PCE Price Index
• How Long Will Inflation Stick Around?

Transcript

Tim Sablik: My guest today is John O'Trakoun, a senior policy economist at the Richmond Fed. John, welcome back to the show.

John O'Trakoun: Thanks, Tim. It's always great to be here.

Sablik: Our topic for conversation today is inflation, specifically how we measure it. Typically, this involves looking at indexes that track price changes across the economy. Listeners are probably familiar with a few of these indexes: the Consumer Price Index or CPI, produced by the BLS, and the Personal Consumption Expenditures Price Index or PCE, produced by the BEA.

In fact, there are many different price indexes, some of them are produced by the Fed or other entities. So, the first question I have for you is, why are there so many measures of inflation?

O'Trakoun: That's a good question, Tim. There are just a lot of different kinds of inflation measures out there, and they all use different strategies to try to get at what the true picture of underlying inflation looks like.

We talk about inflation as the change in the overall cost of living over time. While that's really easy to say, coming up with an inflation measure is a complicated process.

One difficulty is that there are thousands and thousands of prices out there for all the conceivable things we might buy. Collecting information about these prices is a challenge in terms of surveying stores, choosing which stores to look at, or scraping data from the Internet or other sources.

Also, there are some things that are part of the cost of living that are inherently difficult to measure. For example, how do we measure the cost of housing services received by homeowners? It's not as simple as just what these people shell out for their monthly mortgage. That mortgage payment is actually the cost of a financial service. It's not the cost of the housing services that homeowners consume when they benefit from having a roof over their head and a place to sleep at night. Also, what's the best way to measure the cost of health insurance? It's such a complex product — you can't just go into a store and say, "I'd like to buy one unit of health insurance, please."

Assuming we can measure the prices of things we consume, how do we bundle up and summarize all these prices into a single number that represents the cost of living for the whole country? Should we assume everybody consumes the same amount of each kind of good or service? Or should we look at a certain population, like urban residents? Should we track changes in how people spend? Or would it be simpler to just assume people buy the same proportion of the same things every month?

Assuming we can measure prices well and assuming we have a good way of aggregating all these prices into a single cost of living index, how can we know whether a given change in the cost of living at a particular point in time is something that's informative about inflation for the future, versus changes that are just one-time, temporary events? If the cost of living is growing, is it because a single product like gasoline has become more expensive? Or are the prices of many things going up at the same time? And to what extent are those movements telling us something about the future?

These are all really difficult questions that are being grappled with by economists at places like the Bureau of Labor Statistics and the Bureau of Economic Analysis, who calculate the main inflation indexes for the economy, as well as economists at the Fed who use these data to try to figure out what's going on.

Sablik: Thank you very much for that overview. That helps clarify why there are so many different measures.

Inflation metrics are often broken down into headline and core measures, something that our listeners might have heard about. Can you explain the differences between these two, and maybe what are the advantages or disadvantages of each approach?

O'Trakoun: Headline inflation refers to the overall measure that contains all the prices in the basket of goods and services that represents people's cost of living.

As I mentioned earlier, the headline measure can bounce around from month to month for various reasons. So, the increases and decreases in headline inflation in each month are not necessarily informative of what inflation in the future might look like. The top line number might also disguise important information about the balance of forces that affect the overall cost of living, especially when big movements in a single price category can skew the overall number.

Core inflation tries to get a measure more representative of the underlying trend of inflation. And, in theory, it tells you more about where future inflation is headed.

There are many different ways to capture the concept of core inflation. The most popular is inflation that excludes food and energy prices. The idea behind excluding food and energy is that changes in these prices are volatile, so they're less likely to tell you something meaningful about the future.

Another very popular measure of core inflation is the trimmed mean inflation measure. All the spending categories that are included in inflation are ranked according to the size of their price changes. The categories that have the biggest price changes and the smallest price changes are dropped. The surviving categories are then aggregated up to form the trimmed mean measure. This procedure drops the outliers, which are more likely to be driven by one-off, temporary factors that might not be informative about the underlying inflationary environment.

Sablik: What are the sorts of qualities that Fed policymakers are most looking for in an inflation metric?

O'Trakoun: The ideal inflation metric would minimize the amount of noise coming from the inflation data. High frequency changes in that measure would tell us important information about the cost pressures that consumers are facing today and it would give us hints about what inflation might look like tomorrow. Because it would strip out volatile noise from the measure, we would expect a good inflation measure to be smooth, smoother than the headline measure.

In addition, because the ideal inflation metric would be related to the actual price pressures faced by consumers, it would probably also have some relationship with the quantities of goods and services that households are buying. In other words, it would have some kind of consistent relationship with real activity.

Aside from these two qualities, I think a couple of other things might be on a policymaker's wish list. One is that inflation measures should be easy to understand and communicate to the public. Related to that, inflation measures should be as representative of the average consumer as possible, so that ordinary people could look at it and see their own experiences being reflected in the data and see that the Fed is reacting to something that's really related to their experiences.

Sablik: Thinking about all these different measures, you actually recently developed your own new measure of inflation: the trimmed persistence PCE price index. Try saying that five times fast. [Laughs] What's the difference about this measure compared to existing measures?

O'Trakoun: The measure I developed is based off of a topic we discussed on a previous episode of this podcast — link in the show notes.

Sablik: Yep.

O'Trakoun: [Laughs] Which is the persistence of inflation. It turns out that there's a strong relationship between persistence and volatility. When a variable is really persistent, it kind of remembers all the little shocks and jolts that caused it to wiggle around. Instead of letting these shocks fade away over time and going back to some normal level, it hangs on to whatever caused it to move and it's always resetting what it considers to be the new normal. Because of this, a very persistent process is also more variable and volatile.

I use this relationship to make a new measure of inflation that takes out some of the more persistent and volatile price categories. It results in a smoother and less volatile measure of inflation, similar to the way that the most popular core inflation measure takes out the volatile food and energy prices. The difference is that my measure takes out different categories from month to month by updating how persistent each category looks in each month, reconsidering what categories are too persistent and too volatile to make that cut.

Sablik: How do you think your new price index could help Fed policymakers and their decision-making?

O'Trakoun: My new index gives a different take on what underlying inflation might look like in a particular month. It's meant to supplement the other very effective inflation measures that are already out there.

I think the big lesson coming out of the pandemic is how hard it's been to interpret what inflation is doing in real time. For people involved in forecasting the economy, the pandemic has reinforced the need to be humble with our forecasts and our interpretation of the data. Being open to new data that suggests wider possibilities for what economic outcomes might happen in the future, and for the range of policies that might be appropriate, could be part of that quest for humility.

Sablik: You're also very involved in watching what's happening in the economy here at the Richmond Fed Research Department. What will you be watching in the various price indexes as we start 2024 and look into the new year?

O'Trakoun: I'll be continuing to follow the short-term progress in the data to see whether we've achieved a sustained pace of inflation that's at target.

As of today, core PCE inflation has been 2.2 percent annualized over the past three months and 1.9 percent annualized over the past six months. Those are numbers that are pretty much right in line with the Fed's long-run average inflation target. But we'll have to sustain that pace over the coming months in order to get year-over-year inflation that's back down to 2 percent.

Another thing I'll be keeping an eye on is that, in the recent positive news on inflation, we've received help from falling energy prices as well as declining core goods prices. In some months, those declines have offset less positive news coming from core services prices. I'll be watching closely to see how long those positive forces will continue, especially once we start receiving less help from things like supply chain conditions normalizing.

Also, a big question that's out there is do we need to have the composition of inflation look exactly like it did before the pandemic to get back to 2 percent? That's going to be a key question as we keep an eye on the inflation data this year.

Sablik: Well, John, thanks as always, for coming on the show.

O'Trakoun: Thank you, Tim. It's a pleasure to be here.