Comparing Past and Present Inflation – A Comment

The issue

The Consumer Price Index (CPI) published by the BLS has been subject to methodological changes over the course of the years. In particular, because housing is both an investment and a consumption good, until 1983 the BLS included the investment component in the CPI calculation (i.e. “shelter” included mortgage interest rates payments). Therefore, prior to 1983 measurement of shelter inflation was mechanically responsive to the Fed policy via mortgage rates. According to Bolhuis et al. (2022), the BLS methodology made pre-1983 CPI inflation artificially volatile and hard to compare to the current series. (for the record: the current BLS method calculates CPI shelter as “rent of primary residency” and “owner equivalent rent” (OER), therefore it excludes the investment component from the consumers price index)

What the paper does

Bolhuis et al. (2022) construct new historical series for CPI headline and core inflation that are more consistent with current practices for the entirety of the post-war period. Using publicly available Bureau of Labor Statistics (BLS) data, the authors develop new estimates of CPI headline and core inflation that can be better compared across time because they are not subject to methodological changes (that is, they are constructed using the current BLS methods).

Results

Bolhuis et al. (2022) main result is presented in Figure 4 of the paper which we show below. According to the authors, correcting for the change in OER would have reduced the peak of core CPI inflation during the Volcker era by about 5-ish% (the red line in the chart below). The authors argue that the large differences between the official BLS series and their OER-adjusted series reflect both the substantial weight of OER in the index, especially in core CPI, and the lower peaks of estimated OER relative to homeownership costs.

The authors conclude that the current inflation regime is closer to that of the late 1970s than it may at first appear. In particular, the rate of CPI disinflation engineered in the Volcker-era is significantly less when measured using today’s treatment of housing. In order to return to 2 percent core CPI today, the authors conclude that we need nearly the same 5 percentage points of disinflation that Volcker achieved.

A critical comment of Bolhuis et al. (2022)

The exercise of Larry Summers and coauthors is an interesting one (although certainly not new, see for instance Gordon (2003) or Greenlees and McClelland (2010)). However, in our view, it is oversold and its conclusions are a bit stretched for two reasons.

First, the BLS already publishes a methodological consistent series (the so-called CPI-U-RS) that starts in 1978. According to the BLS, at the peak of inflation in June 1980 the official YoY estimate was 13.6%, while the current-method series (the CPI-U-RS) was at 8.8%. Therefore, it is not clear what Bolhuis et al. (2022) contribution is, except for extending the CPI-U-RS series before 1978.

Second (and most importantly), the series constructed by Summers and coauthors is nearly identical to the PCE price index. The reason is pretty simple: the Personal Consumption Expenditure index of the BEA defines the rental value of tenant-occupied housing and the imputed rental value of owner-occupied housing as the amount of money tenants spend for the service of shelter and the amount of money owner occupants would have spent had they been renting. In other words, because PCE measures private spending, the investment component of housing services has never been included in the PCE index (nor consequently in its deflator). Therefore, the PCE price index does not suffer from the same methodological issues of the CPI, at least with respect to the treatment of the OER.

To better understand this point, the figure below compares the BLS core CPI index (the black line), with the BEA core PCE price index (the light blue line), the Bolhuis et al. (2022) alternative index (the red line), and the BLS CPI-U-RS (the green line). All indexes are expressed as year-over-year percent changes from 1968 to 1988. As mentioned by Bolhuis et al. (2022), their indicator (the red line) mimics the BLS CPI-U-RS (although it extends back in time). But most importantly, the red line is nearly identical to the light blue line, the core PCE price index. As we have explained above, this is not a coincidence and comes from the OER item.

Implications for the Fed (staff)

Because the alternative index by Bolhuis et al. (2022) is, in the end, almost identical to the PCE price index (and because the PCE price index is the target of the Federal Reserve), in our view the implications of the authors’ exercise are very limited for the Fed staff and the FOMC. Not only but in our view the authors’ parallel with the ‘70s is a bit exaggerated. For comparison, the figure below shows the YoY of core PCE price inflation from 1960 to present. At the peak, in 1975, core PCE price inflation run well above 10%, while today it is around 5%. Even considering the higher volatility of goods prices (which represented a larger share of the index in the ‘70s-‘80s), the situation today is not as bad as during the Volcker era. This is not to say we have changed our view (or forecast): we continue to believe the current inflation process is persistent (and more than currently perceived by the Fed staff) and that disinflating the US economy requires pushing the federal fund rates well above neutral (markets are slowly realizing it..). But considering the overall picture, including the behavior of inflation expectations, in our view it is safe to conclude that we are not back to the ‘70s. At least not yet.

Contact

Want something more tailored?

We provide tailored consulting on ad-hoc projects.