Thanks to DAve Richardson for his very interesting explanation of the
sources of CPI bias. I have always wondered why the substitution
bias in the CPI is assumed to be positive. It seems that there are
nagative substitution problems of a much greater magnitude. For
example, if large-budget items like housing rise, then families
will have to shift some of their expenditure from other goods
to afford the same level of housing. Given that poor families
now spend as much as 50% and middle-class families as much as
35% of their income on housing, this rise in housing prices
would seem to be an important factor in measuring real incomes.
Similar problems arise with other large-ticket goods like
college education and health care. How does the CPI
account for this effect? Or does the BLS assume that
these effects are offset by the ability to substitute
video-rentals for cable TV, chicken for beef, etc.?
Ellen FRank
