Core CPI in the blogosphere

Over on Michael Froomkin’s blog (Discourse.net), there’s been some discussion of the core CPI and its uses. Michael triggered this by linking to another discussion, in which it was asserted that the CPI left out food and energy prices. I felt I had to chime in, which I did, twice. As reported below.

Comment #1

The CPI does not exclude food and energy prices. Based on the Survey of Consumer Expenditures (the 2005 survey is at http://www.bls.gov/cex/csxann05.pdf), the current weight assigned to food (as a broad category) is 12.8% of the CPI; the weight assigned to energy is more difficult to calculate. However, the weight for "utilities, fuels, and public services" would be 6.9%, and the weight for "gasoline and motor oil" would be 4.3%.

There is a version of the CPI, called the "core" CPI, which does exclude energy and food. Energy and food prices are more volatile--they both rise and fall more than do prices for most other categories of consumer purchases. Since 1960, the rate of inflation has averaged 4.17% per year using the total CPI, and 4.14% per year using the "core" CPI.

It's not clear to me that anyone is doing anything that's nefarious here. How one would construct the charts shown in the post is also something I don't understand, and cannot replicate.

Comment #2

A few more facts. Which this discussion could use.

1. Over the two years (August 2005 to August 2007), the average annual rate of inflation using the overall CPI has been 2.9%.

2. The Bureau of Labor Statistics does not report an annualized rate of inflation for the core CPI (excluding food and energy prices), because it does not think that's an appropriate thing to do. You can calculate it if you wish, but why would you? [If you did, the core CPI (excluding food and energy) over the past two years has averaged 2.5%.] It's designed to abstract from short-run volatility, not to tell a long-run story.
3. The core CPI shows a faster rate of inflation in 11 of the last 24 months. The two indexes show the same rate of inflation 3 times. The total CPI shows a faster rate of inflation 10 times.

4. The core CPI is much less volatile--which is what it's designed to be. The range of annualized monthly rates of inflation using the core CPI varies from a low of -1.8% to a high of +7.9% (the core CPI fell in 3 months, was unchanged in 1 and rose in 20). The rate of inflation measured by the overall CPI has a range from a low of -9.2% to a high of +15.7% (the overall CPI fell in 7 months and rose in 17). But, again, the average annual rate of inflation over the period isn't all that different regardless of which you use. And over an even longer term, they are essentially identical.

5. The BLS also calculates other CPIs, excluding other things. There's one that's all items excluding medical care. And one that's all items excluding shelter. Nobody ever talks about those.

I think if you look at the facts, the conclusion you reach is that the core CPI is not a nefarious means of trying to persuade people that the rate of inflation is lower than it actually is. It's a policy tool that allows a focus on longer-term trends in inflation--and, as a result, the Federal Reserve tends not to obsess about month to-month changes in the rate of inflation. I don't think anyone can argue (successfully) that the Fed has ignored inflation of late.