Ellen wrote, > In calculating > the CPI, the BLS uses fixed weights which are updated only every > decade or so, right?
I'll answer this as best I can--a few years ago I knew all this stuff well but my brain just doesn't remember information like it used to. The answer to the above is: Yes and no. As far as I can recollect the BLS now uses fixed weights for broad categories of goods/services but within these broad categories they take account of the possibility of switches between substitutes. For instance the introduction of the so-called "geometric mean" to calculate prices changes introduces this substitution. They are increasingly building into the price index the switch between substitutions that are more "distant" from each other than before. That is, as far as I can remember they take account of the switch between different types of apples today but they don't consider the possibility that people switch from apples to oranges as the relative price of apples grows. RE >But isn't it > equally likely that it understates inflation, because as prices of > necessities rise (like housing and health care) they eat up a > growing fraction of consumer income and play a bigger role > in the cost of living? I'm not sure about this. The basket of good/services making up the CPI basket is measured kinda-sorta in "real" terms and so as the price of anything increases it raises the total cost to buy the basket. However, an increase in items NOT in the CPI basket (such as certain types of insurance, taxes, the "investment portion" of house payments), which reduces the money left over for the purchase of goods/services in the CPI basket does reduce consumption although the CPI does not take account of this. Eric