>Among other things in OPE-L 5327, I wrote:
>
>>2. For me, it is unclear why the "echo effects will become relatively
>>smaller and smaller over time" if investment as a proportion of capital
>>remains constant or increases over time. My crude examination of the
>>data in the back of Dumenil and Levy tells me that the proportion is
>>more or less constant over time.
>
>In OPE-L 5335, Duncan commented:
>
>The proportion of investment is roughly constant over time, but the whole
>economy grows, so the proportion of investment represented by replacement
>of capital n years back constantly shrinks.
>
>I now add:
>
>If the proportion of investment is constant over time, why isn't the
>proportion earmarked for replacement "roughly" constant over time?
>That is, while I agree there is growth and investment, I'm unclear
>about why there would be a diminishing proportion of gross investment
>designated for replacement.
The proportion of gross investment devoted to replacement of all vintages
is "roughly" constant over time, but that's not what's at issue in the echo
effect, which depends on the propagation of a disturbance of a particular
vintage. Suppose the economy and investment are growing "roughly" steadily
at 50er year, and suppose there was a big downward blip of in investment
equal to 100f GDP 10 years ago. The economy is about 1.65 times as big
now as it was then, so that the blip now is only 10%/1.65 = 60f the
current GDP in terms of replacement. As time passes the blip tends to
disappear relative to the size of the economy, not to propagate. (This
effect is reinforced because replacement is not really bunched, but also
spread out over time, which tends to "blur" the echo even more.) There's no
doubt that you can see some echo effects of particularly large shocks to
investment in the data, but their contribution to the variance at business
cycle frequencies is not very big, I think.
Cheers,
Duncan
Duncan K. Foley
Department of Economics
Barnard College
New York, NY 10027
(212)-854-3790
fax: (212)-854-8947
e-mail: dkf2@columbia.edu