David Levenstam wrote:
>Yes, the BLS series uses CPI-u to deflate the nominal wage series.
>CPI-u doesn't account for changes in the quality of goods or the
market basket,
>and overstates inflation more the higher the actual rate of inflation,
for the
>inflationary period from roughly 1968-1983 the BLS series understates
>wages. Using a better deflator, CPI-x, which accounts for changes in
the market
>basket (though perhaps not for changes in quality) discloses that real
>have indeed risen quite a bit since 1964.

This is completely wrong. The CPI-u is, and the CPI-x was, adjusted for
quality changes (see http://www.bls.gov/cpi/home.htm ). The CPI-X
doesn't exist anymore. It has been replaced by CPI-U-rs (
http://www.bls.gov/cpi/cpirsdc.htm ) which takes all the
methodological improvements introduced in the CPI  in the last few years
and computes what the CPI would have been if those changes had been in
place all along (the published CPI-u for any given date reflects
whatever methodology was in place at the time of the initial release).
The most important change in explaining the difference between the
CPI-u-rs and the CPI-u is the treatment of housing during the late 70s
and early 80s when rents were much lower than the imputed cost of new
housing based on current mortgage rates.

There is a very big question as to whether the CPIs current methods for
dealing with quality change are adequate. On the one hand there are lots
of ways that they probably fail to pick up quality improvements (service
industry output is notoriously difficult to measure so quality
improvements are very hard to discern). On the other hand, there are
clear cases where methods over state quality change. For example, every
time a new product is introduced that replaces an old one the entire
increase in price is attributed to improved quality rather than treated
as a price increase. Since the introduction of a new product is also an
opportunity to change the price, one suspects that at least some
inflation will get factored into such changes, but will be ignored.
Also, a lot of people wonder whether the hedonic methods used to compute
the increases in quality of computers isn't overstating those gains
(which are a very big part of productivity gains over the last decade).
The value of an increase in memory or speed is judged by what those who
buy it are willing to pay for it when both old and new machines are
being sold, but we know that those who run out and buy the fastest new
machine when it comes out probably value speed etc. a lot more than
those who don't buy. When the price goes down enough so that everybody
buys what used to be the fastest machine the CPI quality adjustment
assumes that even the last people to get the fast machine value it as
much as the marginal person when the machine was relatively new.

Most economists who study these issues think that the CPI (even the new
research series) still understates quality change on average though most
also think that the understatement is quite modest (less than the Boskin
commission's estimates of >2%). However, Janet Norwood (who was BLS
commissioner during the last Republican administration) has argued to me
that the CPI probably over estimates quality change for the reasons I
mentioned above. Her guess was that the overestimate was about 1% or
less per year.

Obviously 1% error either way for 30+ years is going to make a huge
difference in what we think has happened to real wages. Given the
methodological uncertainty I doubt we could ever know for sure. - - Bill

William T. Dickens
The Brookings Institution
1775 Massachusetts Avenue, NW
Washington, DC 20036
Phone: (202) 797-6113
FAX:     (202) 797-6181
AOL IM: wtdickens

Reply via email to