[OPE-L:2646] estimation of abstract labor

Fred Moseley (fmoseley@laneta.apc.org)
Sun, 14 Jul 1996 11:07:07 -0700 (PDT)

[ show plain text ]

This is a response to Paul C.'s (2623) and productive and unproductive labor
and the estimation of abstract labor. Paul, thank you very much for your
helpful and open-minded comments.

1. Paul commented:

Thanks very much for your empirical data. The variation in ratios of
productive to unproductive labour is higher than I would have imagined,
and will thus induce more of a bias into our figures than I suspected.


It goes to show that it always worth having more empirical data. When
I said that correcting estimates of labour content for productive /
unproductive labour was the sort of thing that a good graduate student
would do, I was not trying to imply that it was not a worthwhile task,
merely that as professional computer engineer the time that I can devote to
research in economic statistics is more limited. I am sure that in due
course the issues that you raised will be dealt with more thoroughly by
other researchers.

My response:

Until the further work Paul mentions is done, I for one will not have much
confidence in their estimates of abstract labor, nor their results regarding
the correlation between prices and abstract labor.

In addition to the problem of productive and unproductive labor, the other
serious problem with Paul and Allin's estimates of abstract labor (and
everybody else's estimates that I have seen), which we have already
discussed, is the reduction of skilled to unskilled labor and the reduction
of labor of different intensties to labor of average intensity. Paul's
method of using wage rates to reduce skilled to unskilled labor is different
from Marx's concept of abstract labor and makes it difficult to discriminate
between the labor theory of value and the wage-push theory of values.

Until progress is made toward solving these problems, I don't think we can
have much confidence in these empirical results.

2. More broadly, aside from these difficulties, I would ask the question
again (as in 2547): is this the best, or most appropriate, way to
empirically test Marx's theory? I would argue not, because Marx's theory is
primarily an aggregate theory of the broad dynamics of the capitalist
economy and does not necessarily predict a close correlation between
individual prices and individual values. The most appropriate way to
empirically test Marx's theory is to compare the conclusions of this macro
analysis (rising rate of surplus-value, falling rate of profit, inherent
technological change, conflict over the length of the working day and over
the intensity of labor, periodic depressions, concentration of capital,
etc.) with empirical reality. And further to compare the predictive and
explanatory power of Marx's theory with that of rival theory, especially
neoclassical theory, and most especially the neoclassical theory (theories)
of profit or interest.

This is the sort of empirical test of Marx's theory that both Blaug and I
have performed in our recent debate. In the end, Blaug was forced to admit
that Marx's theory does have much more explanatory power than neoclassical
theory with respect to the broad dynamics of capitalist economies. I think
that this is much stronger empirical evidence of the labor theory of value
than Paul and Allin's estimates of the correlation between individual prices
and individual values.