It does not substitute for the unindexed measure, but for "Real GDP" or
"GDP in dollars of constant base-year purchasing power." It corrects for
two biases in constant purchasing power measures, both known to economic
theory for many years. One is the fact that indices of the "Laspeyres" type
tend to overstate.
Suppose some prices go up a lot and others go down a little. Money income
is unchanged, but people can no longer afford the market basket they bought
in the base year; they substitute the now-cheaper goods for the
now-more-expensive ones. But RGDP values these purchases at base year
prices, and we know people could afford the base year market basket at
those prices. Thus RGDP appears bigger than it was, even though people
cannot afford to buy what they could have bought in the base year and may
be worse off in terms of their own prices. In this sense the traditional
measure overstates growth. (This would be clearer if there were any way to
do indifference curves in e-mail).
Moreover, this bias accumulates as the base year prices get more and more
out of date.
The chain index corrects for the latter problem by moving the "base year "
forward along with the "current year," so that each year the index is
computed with the previous year as the base year. That's why "chain." (But
this implies that it cannot be stated in dollars of any year).
It corrects for the first by computing two indices with opposite bias, and
taking the geometric average of them.
I have put a diagram with nominal US GDP and the chain index of real GDP
for 1960-94 at comparable scales on my web server at
http://william-king.www.drexel.edu/top/macro/GDPch/GDP33.html
(I've just been preparing a unit on this for my class, and this rehearsal
is helpful to me! It was all fog to me three weeks ago).
I believe this approach was suggested by Irving Fisher in the
nineteen-oughts as the "ideal index."
Roger A. McCain voice (215) 895 2176
Professor, Economics fax (215) 895 6975
507D Matheson Hall mccainra@dunx1.ocs.drexel.edu
Drexel University http://william-king.www.drexel.edu/
Philadelphia, PA 19104 origin code 507