Slicing and dicing the CPI
admin | May 15, 2020
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.
Collecting price data is especially problematic when huge swaths of the economy are essentially closed. Nonetheless, the Bureau of Labor Statistics slogged ahead and produced the April CPI. Inflation posted disturbingly large declines both for the total and core component. But the price drops were strikingly narrow. An alternative filter, the Cleveland Fed Median CPI, offers a very different view of inflation.
March and April CPI
The CPI in April sank by 0.8%, the sharpest monthly fall since 2008. Energy prices plunged by more than 10%, reflecting a more than 20% seasonally adjusted drop in gasoline prices. Meanwhile, the core CPI decreased by 0.4% last month, the steepest fall on record going back to 1957. Within the core, there were big decreases for hotel rates (8.1%), airfares (15.2%), apparel (4.7%), and motor vehicle insurance (7.2%). The first three relate to sectors where activity was virtually non-existent, while the latter reflects a decision by major insurers to offer rebates to customers since there is so much less driving and, as a result, far fewer accidents and claims. It is debatable whether that decision by auto insurers is a price cut at all, but the BLS viewed it that way. In any case, excluding those four items, the core CPI actually rose by 0.1%
Looking back to March, the story was similar. Headline inflation declined by 0.4%, while the core index slipped by 0.1%. Energy costs dropped by more than 5%, while the core component was driven mainly by three of the same categories cited for April. Hotel rates (-7.7%), airfares (-12.6%), and apparel (2.0%) drove the core component in March into negative territory. Excluding those three categories, the core would have posted a pedestrian 0.2% rise.
Cleveland Fed alternative inflation gauges
The Cleveland Fed for years has produced two alternative measures of CPI inflation that could be thought of as competitors with the core index to offer an accurate view of underlying inflation trends. The core index arbitrarily labels food and energy prices as excessively volatile and focuses on everything else. The Cleveland Fed research effort allows the data each month to determine which line items are outliers.
The first gauge is the median CPI. This concept is simple. Each line item in the CPI is ranked according to the one-month price change and weighted by its importance in the CPI. The line item that includes the 50th percentile of that ranking is the median CPI. The same procedure is reported for 12-month changes as well, to offer an alternative year-over-year gauge. So, the idea is to find the line item that is exactly in the middle, as that should tell us something about general price trends.
The second gauge is similar, a 16% trimmed-mean CPI. Once again, the categories are ranked by price change and weighted by relative importance in the CPI. For the trimmed-mean CPI, however, only the extremes are excluded, in this case, the top and bottom 8% of the index by weighting. Again, the Cleveland Fed produces 1-month and 12-month changes for the trimmed-mean CPI.
The median CPI and 16% trimmed-mean CPI have softened by far less than the headline figures. On a 1-month basis, the median CPI increased by 0.2% in March and 0.1% in April, while the trimmed-mean CPI advanced by 0.2% in March and was flat in April. While the figures were slightly softer than prior trends, they were far from alarming.
Meanwhile, on a 12-month basis, the results are similarly at odds with the more traditional figures. The median CPI decelerated by only one tenth on a year-over-year basis from February to April, inching down from 2.8% to 2.7%. The moderation in the year-over-year trimmed mean CPI was two tenths, from 2.4% to 2.2% (see Exhibit 1). In contrast, the headline and core CPI have cooled by roughly two full percentage points and one full percentage point respectively over the same period.
Exhibit 1: Alternative Year-over-Year CPI Gauges
The March and April CPI figures have been shockingly weak, sparking widespread speculation in the financial markets that the U.S. could be headed toward a prolonged stretch of disinflationary and possibly even deflationary conditions. However, the softness has mainly reflected a handful of categories that have been especially distorted by the response to the virus. Filtering out the outlier categories, inflation has only cooled marginally, at least so far. The medium-term inflation story is still very much up in the air and not yet a convincing sign that the U.S. is in danger of a deflationary spiral.