February 5, 2007
IS MEDIAN CORE CPI A BETTER MEASURE OF INFLATION? (Corrected version)
Note: An earlier version of this story incorrectly identified the median CPI as a top-line index when in fact it's an alternative measure of core inflation. Below, we offer a corrected version of the story and data.
The report on consumer prices for January isn't due for release until February 21, but it's never too early to start thinking about inflation.
Judging by the last update on inflation through December, there's still reason to wonder what's coming. Core consumer prices (which exclude the volatile food and energy prices) advanced by 2.6% last year. That's up from 2.2% in 2005. Will the upside momentum persist?
No one knows, but if you're looking for one more reason to be wary, consider an alternative (and arguably superior) measure of core inflation. The Federal Reserve Bank of Cleveland calculates the median core consumer price index, which is said to minimize temporary "distortions" in measuring price trends that bedevil the conventional core CPI index. The median approach, as an alternative to the core CPI measure, provides a "better" forecast of inflation, the Cleveland Fed opines. "The weighted median CPI is easy to calculate and has a higher correlation with past money growth than other inflation measures, resulting in improved forecasts of future inflation," the bank explains on its web site.
With that in mind, we took the current numbers on the median core CPI and compared them to the conventional CPI. The result: inflation is significantly higher by way of the median benchmark compared to the standard measure issued by the Department of Labor. To be precise, the Cleveland Fed's median core CPI advanced 3.5% last year, vs. 2.6% for the conventional core CPI. (There's a discrepency in some of the Cleveland Fed's median data, but we're using the historical series that shows the latest 12-month change in median CPI at 3.5%, although other portions of the web site quote 3.7%.)
In fact, as our chart below illustrates, the median core inflation rate has been relatively higher in recent history compared to the usual measure. In sum, here's one more reason to delay judgment on whether inflation will be an issue or not in 2007.
Posted by jp at February 5, 2007 11:40 AM
It has always seemed to me that if you are measuring inflation for purposes of setting monetary policy, then use estimations which strip out good-specific effects. In other words, if inflation is too many dollars chasing too few goods, then the impact on prices should be pervasive. Any "average" based method may be biased by large moves in specific goods, which are likely due to conditions in a specific goods market as opposed to anything related to monetary policy.
Posted by: Tom Graff at February 5, 2007 1:08 PM