CPI and all that

[enilsson@xxxxxxxxx]

Re the discussion of CPI.

I wrote about some of the issues related to a "cost of living" index and a CPI
in a 1999 Review of Radical Political Economics article. I think I have my own
best estimate of a cost of living index in it.

The CPI is almost always used -- wrongly -- to generate "real" values,
particularly for wages and earnings.

Among the problems: the CPI assumes constant preferences -- in short it is
assumed that folks in 1947 have the same likes/dislikes as folks in 2000. This
is obviously false and preferences have become, if anything, "more demanding."

The CPI is the cost of a "market basket" but it ignores the impacts on
standards of living caused by increased housing costs, changes in the relative
size of interest payments, changes in the tax bite, and so on.

The CPI can't be used to deflate benefits as the "market basket" does not
include benefit like products.

The CPI-U-X1 has its flaws. It likely takes out too much of the costs related
to housing than is appropriate (as recognized by others). Changes in the asset
component of housing are excluded from the CPI not because they are unimportant
but, simply, because the CPI has been defined to exclude assets. There is no
other reason then this.

For what it is worth, the work of the "Boskin Commission" was amazingly shoddy
on almost all counts. I don't think anyone of those on the committee really
know what the CPI is.

Eric