Of course, CPI is an average and no individual actually buys exactly the basket of goods in CPI, so some caution is in order. The sources of bias cited in the paper can perhaps help one in making an adjustment:
- commodity substitution bias occurs, for example, when a real consumer notices a jump in the price of beef and starts buying pork instead (Michael James noticed a "swine flu special" on pork) but CPI continues to track beef the amount of beef consumed exactly as before. If you think pork and beef are not really equivalent, that such substitution is in effect a loss of real value, then subtract 0.15% from the 0.6% number above.
- outlet substitution happens when you buy the same designer jeans at lower price at the outlet store instead of the high-priced retail store with the loud music. CPI takes a while to catch up to this shift retail patterns. Think part of the pleasure of the jeans is the music? Take away another 0.1%.
- quality change bias might best be characterized by considering quality of the average car of 25 years ago compared to today. I for one prefer today's cars. If you really enjoyed the breakdowns and rust of bygone days, you can say inflation is not really lower by another 0.15% compared to CPI.
- new goods, new services and new brands bias happens when the increased in standard of living resulting from the introduction of such new products and greater choices is not reflected in CPI. If you think such things as microwaves and cell phones have not enhanced your life, then take away 0.2%
Overall, I am most skeptical about the first bias, commodity substitution, since when I am forced by price to buy downward, it isn't the same value, though I can imagine other substitutions where I would not care. The other biases do ring true enough, so I'm happy to accept that CPI really does over-state inflation.