>A few comments on Duncan's OPE-L 4823.
>
>Duncan states:
>
>The "r" method will coincide with the RRI if the depreciation schedule used
>by the capitalist accountant turns out to be the true economic depreciation
>ex post, which it usually isn't.
>
>John comments:
>
>I agree that this is true for the individual capitalist. But the "r" method
>is generally used for all capitals in a given period. That is, we look
>at the average "r" for a given year. Even abstracting from technical change,
>changes in this annually computed "r" will occur as the stratification of
>fixed capital changes. Or, as the average age of the fixed capital decreases,
>this annual "r" would fall as the RRI remains constant and as it increases
>the annual "r" would increase. I hope this is clear. Since in the
>latter part of your post, you expressed possible disagreement on this point,
>let me see if I can be a bit clearer.
>
>If I invest in a machine and expect a certain return each year, which includes
>both profit and depreciation, then as I collect the depreciation funds year
>after year my "r" for the latter years increases. Why? I have less invested.
>Clarity?
Well it's true for individual investments, but maybe the averaged data
isn't so bad as a representative of the individual investments. Firms
typically invest fairly steadily over time, so the average age of their
_total_ capital doesn't decline. It does fluctuate according to their time
profile of investment, but that effect is relatively minor quantitatively.
Thus the total of their profits divided by the total reproduction cost of
their capital will often track the RRI pretty closely. Anyway, that's the
assumption I think everyone implicitly makes in using macro data to study
the issue.
..
>I was trying to raise the
>issue of how one can transform values into prices of production without
>prior knowledge of prices of production. That is, how do we know how
>long fixed capital will last on average if all we know are the embodied
>labor values. Clearly, this is not simply technical data. Rather
>fixed capital is put out of its misery by its own lack of profitability.
>But if we compute the "RRI" (or for that matter "r") based upon
>embodied labor values, we generally get a different rate of profit than
>that based upon prices of production. The lifetime of fixed capital
>thus cannot be known without knowing prices of production. Hence,
>the idea that both "c" and "v" are known PoP's ex ante as they are
>in Marx's transformation procedure is not a mistake on his part but
>rather a recognition that this was the only way to proceed with
>the analysis.
I think this is correct in theory. In a competitive system a capital good
is retired when the value of its output falls below the wage required to
operate it (either because it breaks down or wears out, or because
technical change has lowered the price of the output relative to the wage).
Thus if one were to use embodied labor coefficients (direct prices) in
general the retirement point will be different than for profit-rate
equalizing prices. Thus in principle the perfect foresight depreciation
schedule for direct prices will be different from the perfect foresight
depreciation schedule for prices of production.
But we know that depreciation accounts are very rough estimates of real
economic depreciation to begin with. They probably don't have much real
information at the aggregate level about economic depreciation, and to the
extent they do it is probably because large errors in individual accounts
to some degree offset each other. I wonder whether what we know to be
relatively minor differences between direct prices and prices of production
would lead to very big differences in economic lifetime calculations.
In any case, what matters is what capitalists actually do, and that is
presumably based on their observations of real market prices and their
projections based on these, not on direct prices, which they can't observe
and would have no interest in.
>
>
>John concludes:
>
>Given the differences between "r" and "RRI", it is hardly surprising
>that Dumenil and Levy find a lag between the two over time. Why both
>are falling, albeit lagged, requires some explanation. Here again, I
>would find things much more convincing if the analysis itself contained
>examples of how individual capitalists were actually investing. As it
>stands, the reader is left inventing scenarios of such investments.
I know you feel strongly that capital-using investments are rare at the
micro level. It is true that at the macro level a capital-using,
labor-saving bias in
technical change is quite common (though by no means universal). The paper
I wrote this winter with Adalmir Marquetti, which is on my web site
(www.columbia.edu/~dkf2) looks at the Penn World Tables data on 126
countries over the 1960-1991 period. Over 500f the country-year
observations have this pattern.
Maybe there's more capital-using technical change in sectors that you don't
observe, or maybe there's some bias in the way the capital value data is
constructed, or maybe the macro data patterns have more to do with changes
in relative prices of capital goods to wage goods.
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