Selection Bias in the Consumer Price Index?: The Ticker

a | A

By Caroline Baum

When statisticians use the term "sample selection bias," they are referring to a flaw in the selection process that influences the outcome of a study and produces distorted results.

That phrase came to mind just now when I turned on the radio and heard the news anchor say that consumer prices were unchanged last month after falling by 0.1 percentage point in October. In both cases, a decline in energy prices offset increases elsewhere.

The core consumer price index, which excludes food and energy and is touted as a truer measure of inflation, rose 0.2 percent last month. The 2.2 percent year-over-year increase is the biggest since October 2008. Just take a look at the graph: Which way is the trend?

When energy prices boost headline inflation in any given month, the commentary focuses on the usually tame core to support the conclusion that there is no inflation. When energy prices depress the CPI, somehow the focus shifts to the headline CPI -- once again to show there is no inflation. That's what made me think of selection bias, at least in terms of what we choose to see.

The Federal Reserve has an implicit inflation target of 1.5 percent to 2 percent. With the CPI up 3.4 percent in the past year and the core up 2.2 percent and climbing, policymakers better hope those inflation expectations are well anchored.

(Caroline Baum is a Bloomberg View columnist.)

-0- Dec/16/2011 18:52 GMT