Unraveling an inflation puzzle

Stephen Stanley | December 6, 2024

This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors. This material does not constitute research.

Fed Chair Powell has set the tone for the FOMC inflation narrative for several years.  In late 2022, he introduced a new framework for assessing core inflation, consisting of three components: core goods, housing services, and core services ex-housing.  More recently, he has argued that all but the housing component have posted inflation rates consistent with the Fed’s 2% target, which seems to be inaccurate in the case of core services ex-housing.  Recent comments by Powell and New York Fed President Williams clarified what policymakers have in mind.  However, an alternative method of parsing the inflation data, the Atlanta Fed’s Sticky-Price CPI gauge, suggests that Fed officials may be overly upbeat in their near-term outlook for prices.

Core inflation as the Fed sees It

In November 2022, Powell introduced a new framework for core inflation, dividing the aggregate into three pieces: core goods, shelter costs, and core services ex-housing.  At that time, core goods inflation had returned to pre-Covid levels, as the supply chain difficulties created by the pandemic had dissipated.  Housing costs were rising at an elevated pace, but Powell assured that they would come down relatively soon, as shelter costs tend to respond to swings in home prices and market rents, which had already begun to cool, with a considerable lag.  That left the Fed’s focus on bringing down the inflation rate for core services ex-housing.  Financial market participants also began to concentrate on this measure, which was dubbed the “supercore.”  At that time, the “supercore” was rising at a 5% year-over-year pace.

Two years later, not much has changed for the first two components.  Core goods prices are still pretty close to stable, not far from where they were before the pandemic.  Shelter costs have moderated to a degree but remain too high, and Powell continues to argue that the lags will kick in soon. I am growing skeptical on this front, but this piece will focus elsewhere.

That leaves the “supercore,” for which inflation has decelerated to 3.2% on a year-over-year basis through October, an improvement but still well above the Fed’s 2% inflation target.  I have been puzzled in recent months by Powell’s repeated statement that the core services ex-housing component has “returned to rates closer to those consistent with our goals” (taken from Powell’s November 14 speech).  In fact, prior to the pandemic, this measure ran just above 2%, about a percentage point lower than the current pace (admittedly, overall core inflation ran slightly below 2% during that period).

However, Powell explained at his November 7 press conference and New York and Fed President Williams noted in his speech on December 2 that the Fed is actually looking further back to the period between 2002 and 2007, which, according to Powell, was “the last time we had sustained 2 percent inflation.”

It is interesting that the FOMC would skip over the 2016-2019 period, when core inflation generally ran below 2% but only by a few tenths of a percentage point and was relatively stable, to an era much further back in history.  There are two features of that earlier period that make it, in my view, a less apt comparison, aside from the obvious one that the structure of the economy today is far different than it was 20 years ago. First, while core inflation may have averaged 2% over the period from 2002 through 2007, it was not exactly steady, as Powell suggested.  In fact, it was accelerating throughout that period.  The December-over-December annual gains for the core PCE were, beginning in 2002: 1.8%, 1.6%, 2.1%, 2.3%, 2.3%, and 2.4%.  It is true that core services prices ex-housing were running at a relatively steady pace of inflation of between 3% and 3.5% during that period, but the low core inflation performance in 2002 and 2003 was driven by steep declines in core goods prices (-1.3% for December/December in 2002 and -2.3% for 2003), a feature that was not sustained in subsequent years and is unlikely to be replicated in the near future.

More broadly, during that time, the FOMC was concerned about the prospect of excessively low inflation in 2002 and 2003, fearing that the US would fall victim to the deflationary spiral suffered by Japan at that time.  However, extremely easy monetary policy eventually boosted the economy sharply as well as boosting inflation, forcing the Fed to abruptly shift its focus in 2004.  The FOMC hiked rates from 2004 through 2006 to fend off a steep acceleration in prices and then kept the policy rate restrictive throughout 2007.  Even in 2008, after the financial system began to unravel, the FOMC maintained a tight policy stance for much of the year because officials feared that inflation would take off.  Thus, labeling that period stable in terms of inflation seems like an inappropriate characterization.

Second, the inflation dynamics during that period were driven to a significant degree by upward trending energy prices, a very different inflation environment than the Fed is attempting to manage now.  Thus, I find the choice of the 2002-2007 period as the benchmark for core inflation as a little puzzling.  Perhaps, Powell and the FOMC are revealing, whether intentionally or not, that they would be more comfortable to see core inflation run in the 2.3% to 2.4% range seen in 2005-2007 than the 1.8% average pace seen in 2018-2019.

Supercore is not so super

As I have pointed out in a number of pieces, I do not view Powell’s 3-component framework for assessing core inflation as ideal.  Just as it is arbitrary to exclude food and energy prices from the inflation numbers to calculate “core,” it seems arbitrary to pull shelter costs out to create a “supercore” gauge.

Instead, I see the key distinction in sifting through the core inflation components to be the degree of short-term volatility in the various line items.  Certain categories tend to exhibit tremendous seasonal volatility such as apparel, while others routinely rise and fall substantially, sometimes having little or nothing to do with broad economic fundamentals, such as used vehicle prices and airfares.  This is why in my analysis of the core CPI over the years, I have isolated the five noisiest categories, which I have dubbed The Fearsome Five, and I focus on core inflation excluding those line items.

For what it is worth, the “ex-Fearsome Five” core CPI gauge rose on average by 2.13% per year during the 2002-2007 period and by 2.24% in 2018-2019.  In contrast, this aggregate surged by 3.9% last year and is on pace to advance by about 3.4% in 2024.  This suggests that the underlying trend pace of core inflation remains much higher than is likely to be consistent over time with 2% inflation, whether the benchmark is 2002-2007 or 2018-2019.  To be clear, I do expect further moderation in core inflation going forward, but this measure reveals that it is far from the case that everything but shelter costs is already roughly where it needs to be.

The Sticky-Price CPI

The Atlanta Fed has crafted another alternative gauge of underlying inflation called the Sticky-Price CPI.  I have written about it in the past, but a quick reminder is in order.

Economists have studied the frequency at which different prices change.  One academic research effort found that of the 350 detailed categories used in the CPI, the median frequency of price changes was once every 4.3 months.  Fed researchers built on this paper by dividing the 45 major components of the CPI into the 21 with more frequent changes than 4.3 months and the 24 with less frequent changes.

The former grouping was labeled the “Flexible-Price CPI” and includes such categories as most of the food and energy sectors, used vehicle prices, hotel rates, and several apparel components.  In theory, the Flexible-Price CPI should be quite sensitive to the state of the economy (that is, prices for these items closely track frequent shifts in supply and/or demand).

The latter group was termed the “Sticky-Price CPI” and includes categories like rent, OER, medical care services, motor vehicle insurance, household furnishings and operations, and recreation.  In theory, because firms selling “sticky” products and services move their prices infrequently, in many cases because the adjustment costs of altering prices may be high, the sticky price index tends to be more forward-looking.  That is, firms will want their prices to account for inflation over the period between the infrequent price changes.  Thus, the Sticky-Price CPI can be thought of as offering particular insight into inflation expectations and the underlying rate of inflation.

I like the Sticky-Price CPI for another reason.  The sticky-price components typically exhibit less random noise and thus offer a more accurate view of the underlying trend when some of the “flexible” line items (like used vehicle prices or airfares) are swinging the overall core figure up and down.  In this sense, the Sticky-Price CPI approximates my Ex-Fearsome-Five core CPI measure.

Flexible prices

The Atlanta Fed Flexible-Price CPI actually correlates fairly well with the core goods component of Powell’s three core inflation groups. The Flexible-Price CPI jumped for a time during and just after the pandemic and then returned to the flat-to-slightly-negative trend seen in the years prior to Covid (Exhibit 1).

Exhibit 1: Atlanta Fed Flexible-Price CPI

Source: Atlanta Fed.

Sticky prices

In contrast to flexible prices, the Sticky-Price CPI has decelerated but remains elevated. The year-over-year advance in this measure has declined from its peak but is still running at about 4%, far too high for the Fed to take comfort (Exhibit 2).  This gauge resided in a 2% to 3% range in the years prior to the pandemic.  For reference, the sticky-price CPI ran in the 2% to 2.5% range for the early part of the 2002-2007 period before accelerating to around 3% in late 2006 and 2007.

Exhibit 2: Atlanta Fed Sticky-Price CPI

Source: Atlanta Fed.

One might take comfort from the decline in the Sticky Price CPI by noting that the year-over-year advance has been steadily coming down, a trend that continued in October.  However, a more granular examination of the Sticky-Price CPI is less favorable. The 3-month annualized percent change, a more sensitive way of slicing the data, shows the rate of inflation sank to a low of 2.7% annualized in July, which was the last reading that officials would have seen in the weeks leading up to the September FOMC meeting.  One can understand why the Committee was willing to get aggressive and cut by 50 bp at that meeting.  However, since then, the gauge has accelerated for three straight months and is back to 3.7%, not far below the year-over-year advance and well above anything seen in either the 2002-2007 or 2018-2019 eras.

Exhibit 3: Atlanta Fed Sticky-Price CPI—a closer look

Source: Atlanta Fed.

Conclusion

Powell, Williams and others on the FOMC have laid out a narrative that core inflation with the exception of housing has returned to levels broadly consistent with the Fed’s 2% target over time.  However, in my view this interpretation of the data reveals the flaws in the framework that Powell and others have chosen to employ to assess the data.  Looking at other gauges of underlying inflation, like the Atlanta Fed Sticky-Price CPI index—or my own ex-Fearsome Five core CPI measure—suggests that the Fed has considerable work still to accomplish to get back to 2% inflation.  Moreover, the uptick in core inflation over the past few months suggests that the risk is rising that, as Governor Waller noted on Monday, “progress on inflation may be stalling at a level meaningfully above 2 percent.”