Related Links

• Relative Price Changes Are Unlikely to Account for Recent High Inflation
• How Broad-Based Is the Recent High Inflation?
• Forecasting Inflation

Transcript

Tim Sablik: Hello, and welcome to Speaking of the Economy. I'm Tim Sablik, a senior economics writer at the Richmond Fed. I'm joined today by Alex Wolman, vice president for monetary and macroeconomic research in the Research Department at the Richmond Fed. He's here to talk to us today about his recent work, exploring what's been happening with inflation lately.

Alex, welcome.

Alexander Wolman: Thanks a lot, Tim. It's great to be here.

Sablik: Great to have you.

I think a lot of us have been hearing a lot about inflation over the last few months. It's been rising at a pace not seen in decades. Could you start by explaining how the Fed thinks about inflation, which might be a bit different from how the average consumer thinks about it?

Wolman: Sure, I think people tend to view inflation in different ways if they're not following the official data on inflation closely. I think it's common for people to think about inflation in terms of specific price increases that are most visible to them. For many people, food and gas prices are most visible, but for some it could be other prices that might be related to their work. For example, someone in the construction business might associate inflation with large movements in lumber prices.

For economists and for the Fed, inflation is generally defined as the average rate of price change for consumer expenditures. It's really a double average in that it averages over all consumers and over all expenditure items.

My colleagues, Claudia Macaluso and Felipe Schwartzman, have been looking at how inflation differs across people because different people have different consumption baskets. My focus is on the variation in price changes across expenditure items and how changes in the distribution of price changes may be useful for interpreting the behavior of overall inflation.

I want to emphasize the big picture question up front — which we'll be coming back to — which is how to interpret inflation when it comes in far from the Fed's target. We'll be going into a lot of details, but that's why we're doing this.

Sablik: Yeah, that's a good point because the focus of your research is really trying to determine what all these movements mean in terms of the Fed's target and monetary policy.

When you're examining these different components of inflation, how does that differ from other measures that people might be familiar with like the Consumer Price Index, CPI, or the Personal Consumption Expenditure Price Index, PCE?

Wolman: Yeah, PCE inflation, I'd like to focus on that because that's really the most relevant for our discussion.

Roughly speaking, PCE inflation is a weighted average of the price changes for all categories of consumption. I say a weighted average [because] the weights on each category are the entire economy's share of expenditure for those items. For example, if consumption expenditures [are] made up of 2 percent gasoline expenditures, then gasoline would have a 2 percent weight, approximately, in the PCE inflation rate. What I've been focused on is the distribution of those price changes for the different categories of consumption and, more specifically, the distribution of relative price changes.

Now we're going to start getting into some details, so let's see how it goes.

Sablik: Alright.

Wolman: A relative price change is the difference between an observed dollar price change and the inflation rate. For example, if the inflation rate is 2 percent and the dollar price of light bulbs rises by 1 percent, then there's been a 1 percent decrease in the relative price of light bulbs. If the dollar price of eggs rises by 5 percent when inflation is 2 percent, then there's been a 3 percent increase in the relative price of eggs. So, that's a definition of relative price changes.

Just from that definition, there are a couple of interesting facts that follow directly. The first one is that, across all expenditure items, the average relative price change has to be zero. Now I think this is pretty easy to understand. The relative price change for any one item, say light bulbs, is the difference between its dollar price change and inflation. Inflation is the average dollar price change. So, the average relative price change has to be zero.

Simple?

Sablik: Maybe. Let's see if I got this right. When we're thinking about a measure of inflation like PCE, that is the average price change across all these different categories. Then, if you take the average relative price change of all of those, the average of the relative price changes is zero because they're kind of going to cancel out.

Wolman: Yeah, exactly. For each individual item, you have this meaningful thing which is the difference between the price change of that item and inflation. But, just as you said, inflation is the average of all the price changes. So, once you just do the math and take the average of those relative price changes, it's got to be zero.

Sablik: Okay. We're getting to why that's potentially interesting. You mentioned there's a second fact that kind of jumps out.

Wolman: Yeah, there is. The second fact is not as obvious — there's no necessary relationship between the distribution of relative price changes and the level of inflation. Take one extreme where every relative price change is zero. That means that every item has the same price change as the inflation rate, the prices are changing at the inflation rate for every item. Another extreme [is] there could be a large share of small relative price increases and a small share of large relative price decreases, or vice versa. All those different scenarios can happen with the same level of inflation.

So again, there's no necessary relationship between what's going on with the distribution of relative price changes — it has to average zero — and the level of inflation. So, I want to establish that.

Sablik: Okay.

Wolman: Now, I just told you there's no necessary relationship between the distribution of relative price changes and inflation. They might have nothing to do with each other. But the work I'm doing is actually all about investigating that relationship. There doesn't have to be relationship, but when you look at the data there's some interesting stuff there.

I'm focusing on one statistic in particular, which is the share of relative price increases in the economy. I'm looking at how that share has correlated over time with the inflation rate, with the goal of helping us to interpret the behavior of inflation.

Sablik: Okay.

Wolman: I realized I've been going into some detail here, but I'd like to actually dig a little bit deeper on this statistic that I mentioned — the share of relative price changes. We talked a minute ago about how the average relative price change has to be zero. But if I can indulge in a little "podcast algebra" on our imaginary blackboard here.

Sablik: Alright, everybody at home, get out your whiteboards.

Wolman: Probably should not draw on these beautiful white walls.

We agree that the average relative price change has to be zero. From that fact, it follows that if there's a large share of relative price increases, say greater than one half, the average size of those relative price increases has to be small compared to the average size of relative price decreases. Likewise, if there's a small share of relative price increases less than a half, then it has to be that the average size of those relative price increases is large compared to the average size of the relative price decreases.

Sablik: Okay, so another way of thinking about that is if we look at the relative price increases, if there's a lot of categories of consumption that are increasing at a faster rate than the rate of inflation, PCE, then it has to be the case that if many categories are above that rate then the smaller number of categories that are below that rate are much more below to kind of balance things out.

Wolman: Exactly.

Sablik: And vice versa.

Wolman: Exactly. That was just perfect, so perfect that I want to repeat it. If there are a lot of categories that are increasing prices faster than the inflation rate, then they can't be increasing very much faster than inflation. Otherwise, the inflation rate would be higher. As you said, we know inflation is what it is, so then it's got to be that there's some other prices that are either falling a lot or are rising much less than inflation.

Sablik: Yeah. Maybe we can pause here. This work has a connection to something that our listeners might be familiar with, which is this idea of core inflation, right, which is another way of getting at this.

Wolman: Yeah, yeah. It's important that I should mention that, so thank you.

Core inflation is just the overall inflation rate without including food and energy prices. Trimmed mean inflation, which many economists look at but is less well known, is similar. But instead of omitting those two categories, it omits a broader set of categories that are in the extremes of the price change distribution.

Those are popular measures that are also meant to help us in interpreting the behavior of overall inflation. But they come at the issue from a little bit different direction than what I'm doing in that they adjust the inflation rate to come up with a different inflation measure. What I'm doing is taking the existing inflation measure and using information about the distribution of relative price changes to help in interpreting it.

Sablik: Great. Now that we've laid that groundwork, when you took this approach and looked at historical data, is there some sort of pattern that emerges, particularly if we think about relative price changes since the mid-90s when the overall rate of inflation was generally low and stable? What did you find there?

Wolman: Well, I told you about the statistic — the share of relative price increases — that I'm focused on. It turns out that over that period, from 1995 through 2019, inflation was close to the Fed's 2 percent target and it was fairly stable. (The 2 percent target didn't become official until 2012, I should say.) But over that period, there was actually a tight decreasing relationship between the share of relative price increases and the level of inflation. Keeping in mind that inflation was fundamentally stable, but nonetheless, from month to month there were a sizable variation in inflation.

Turns out that in months when inflation was relatively high, the share of relative price increases was low [and] the average relative price increase was high. Loosely speaking, we can say that high inflation in months when it happened was accounted for by a small fraction of especially large price increases. Similarly, low inflation, when it occurred, was accounted for by a small fraction of especially large price decreases.

Sablik: Right. When we think about the current run-up in inflation, when it began last spring it seemed to be mostly limited to a few sectors such as lumber and automobiles.

Wolman: Yep. In those first several months of the current high inflation episode, the data seemed to be following that same pattern that I described. High inflation was occurring, but it corresponded to a small share of large relative price increases, consistent with the stories that were all over the place at the time about astronomical increases in used car prices, for example.

Sablik: Right. I guess the next question I have is, from the perspective of the Fed, does it matter if higher inflation is happening across a bunch of consumption categories in the economy versus just a few places?

Wolman: Well, when we see inflation much higher than recent history – than that stable period from 1995 to 2019 – we see inflation much higher than the Fed's 2 percent target. The Fed policymakers are then faced with the question of whether that high inflation is being driven by unusual supply and demand factors for particular goods and services against a background of fundamentally stable and effective monetary policy. In that case, inflation would return close to target when whatever those unusual factors are are resolved. Or does high inflation represent to some degree unusual behavior of monetary policy? If it's the latter, then the Fed would need a plan for policy to return to normal.

If there are big price changes in just a few categories … while others are behaving as in the past, then it's more likely that higher inflation is actually being caused by unusual supply and demand factors in those categories. There's little that the Fed can or should do about those sectoral factors. In contrast, if high inflation is more broad based, then it's more likely associated with monetary policy and would be responsive to monetary policy.

Sablik: Gotcha. Keeping that in mind, you mentioned that from the spring of 2021 until around October of 2021, inflation was following this historical pattern that you'd seen in the '90s. What changed since the fall of last year?

Wolman: In the last four months of data that we have — we have PCE inflation data through January — the data have deviated from that script. Not only has inflation continued to be high, but it's been quite a bit higher than we would have predicted, based on the share of relative price increases using that relationship that jumps out of the data from 1995 to 2019. In January, which as I said is the most recent month for which PCE inflation data is available, the share of relative price increases was very close to one half, which historically corresponded to an annualized inflation rate of around 2.5 percent. Actual inflation in January was 7 percent at an annual rate.

Sablik: Oh, wow. So, let me see if I got that right. That means about half of the goods in the economy were increasing at a rate faster than the inflation rate. Based on history, we would [have] predicted that that would have resulted in the overall inflation rate about 2.5 percent. Instead, we're seeing 7 percent which is much higher.

Being in those uncharted waters, what implication does it have for the Fed in terms of monetary policy?

Wolman: When we look at the period from 1995 to 2019, we have the benefit of hindsight. We know that whatever else was going on in that period — and there was a lot going on — inflation was quite stable by historical standards and it was quite close to the Fed's 2 percent target. (Again, that became the official target in 2012.) In months when inflation moved far above target, it was associated with a small share of large relative price increases.

Now we've experienced several months away from that close relationship, and I need to answer your question. The fact that high inflation has been associated with a larger share of increasing relative prices suggests, as I indicated earlier, that expansionary monetary policy is an important factor in explaining the recent high inflation numbers. I'm pretty confident that that's the case.

But expansionary monetary policy itself can come in two forms, and this is very important. It can represent temporary shocks or temporary deviations within a stable monetary policy regime, or it can represent a change in the monetary policy regime. In the former case, where it's temporary shocks and a stable regime, if monetary policy simply returns to its normal policy, then inflation will follow suit, not immediately but soon. But in the latter case, that would correspond to a un-anchoring of inflation and the Fed would need to take stronger actions in order to re-anchor inflation.

I want to talk about how I see the current situation relative to those two possibilities. It seems clear from FOMC communications that there's no intended change in regime. The FOMC has a stated target of 2 percent inflation and the latest projections — the summary of economic projections from FOMC members — indicate that they expect to conduct policy in order to return inflation to the 2 percent target. In addition, we have information about what markets think about inflation from nominal and inflation-indexed long-term Treasuries. Those measures of inflation compensation suggest that financial markets don't perceive a change in regime.

So, I do think there's a message for the Fed in this data. It looks to me that policymakers are responding to that message with a clear plan to normalize policy that is credible for markets.

Sablik: That sounds great, but we both work for the Fed. What about people that are listening outside who maybe are less optimistic about inflation and are thinking back to the experience of the 1970s?

Wolman: I want to stress one fundamental difference between the current monetary policy environment and the high inflation period of the 1970s. Today, the Fed has a stated target for inflation of 2 percent. Four times a year, members of the FOMC publish their own individual projections — forecasts for the economy and for what they think is the appropriate policy path that's going to be consistent with their forecast. They're providing information about how they expect to attain the inflation target over the medium run. Nothing like that existed in the high inflation period of the '70s to anchor inflation. Inflation was just a moving target.

Sablik: Yeah and, as you say, there's so much more information out there, including the information that you're sharing with us on this podcast. We definitely appreciate you coming on to talk about it.

For our listeners that are interested in following more of your work on this, they can head over to our website at Richmondfed.org. You've got a couple of Economic Briefs [in September 2021 and March 2022] on this topic and some other work, so I'm sure people that are interested in following this can check it out there.

Wolman: Thanks a lot, Tim. In addition to the Economic Briefs, I have a paper in the works in this area. And now I'm saying that on the air, so I better get it right. Thanks a lot.

Sablik: Thanks for being here.