Two major methodology changes were introduced since the early 1980's. First, the CPI has been disconnected from real estate prices. This methodology change helped to inflate the mortgage bubble and the subsequent crisis. Secondly, the “hedonic adjustment” was introduced in the 1995 by the Boskin commission. The CPI has no longer been a proper measure of inflation since then.
What should we use instead? There is a service called Shadow Government Statistics, but it is not reliable. (It's good for conspiracy theorists, but not for serious economists.)
Is there a good alternative to CPI and ShadowStats?
I made an experiment recently. I downloaded a couple of inflation data series from the the FRED database: Food, Fuel oil and other fuels, Medical Care, New vehicles, Services, Transportation, and Finished Consumer Goods. Then I calculated a simple unweighted average out of the seven series, and compared the outcome with the official Consumer Price Index for All Urban Consumers (All Items.)
I was stunned with what I saw:
Now this is the big picture:
The 1.2% annual difference may look innocuous at first, but nevertheless, it may be the kind of difference that might cause serious monetary policy mistakes. It's likely it has already happened.
The official CPI figures are bogus indeed. From now on, however, we have the Devil's Price Index. Accept no substitutes.