RE: Those questionable productivity numbers

[enilsson@xxxxxxxxx]

This is primarily to Doug.

I was going to respond to Doug's question of the other day but I'm still daddy
at home and don't have Doug's message. But, I'll give a shot at it anyway.

The question was something like why can't one compare a change in the absolute
value of a nominal variable with the change in the absolute value of a real
variable.

I'm not sure that my answer is understandable but this is due to a lack of
clarity on my part rather than the reader's problem.

First, one can do this and it might say something interesting, like Doug's
observation linking changes in nominal computer spending with real computer
spending.

Second, this comparision might be very misleading.

Real series are not intended to be used in comparisions of LEVELS but for
comparisons in PERCENTAGE changes.

The level of a real series depends on the year that is used to provide the base
year prices. If, say, the BEA has used 1999 as the base year then 1999 real and
nominal computer spending would have equalled. In this case, the real growth in
computer spending would have been very close to the nominal increase and not
the big differences seen in the actual numbers.

That is, the size of the growth of the real variable depends on the base year
chosen which is arbitrary. A change in the base year prices will not, however,
lead to a change in the percentage growth in the real variable.

(The details are really more complicated because changing the base year also
changes the relative importance of different sectors of the economy, etc).

The effect of a given year's increase in nominal spending on computers on real
spending on computers depends on the base year used to generate the real values.

The big difference in the nominal spending values on computers and the real
spending on computers is really a measure that the computer sector is "really
larger" today than the nominal values give it credit for.

Actually this last point might be important. But I'll add to this in my next
message. And, maybe I'll be more coherent.

Eric