[OPE-L:5510] Re: Humbug Aggregate Price-Value Correlations

Tsoulfidis Lefteris (lefteris@macedonia.uom.gr)
Tue, 23 Sep 1997 07:20:06 -0700 (PDT)

[ show plain text ]

>From the recent exchange of views on whether one should take seriously
or not the correlation coefficients as summary statistics of the
closseness of labor values (or prices of production) to market prices,
it is clear that one cannot rely on cross section regressions between
interindusty values and market prices because of the problem of spurious
correlation. This issue is well known from the very first studies of
Shaikh and Ochoa and this is the reason why a series of other summary
statistics (MAD, MAWD, Vector norm) are suggested in an effort to
quantify the true magnitude of the deviation of labor values or prices
of production from market prices.

In these studies a roughly 20 0eviation measured by MAD or MAD weighted
by the output of each sector are in the range of acceptance of the
proposition that labor values or prices of production are close enough
to the observed prices. These are the results reported for Italy in the
Varzi study mentioned in Shaikh (1983) and for the US economy the MAD is
significantly less than 20 0n the recent years 1967, 1972 but the MAD
gets closer to 20 0n the first years of the study 1948, 1958, 1963.

My joint (with Th. Maniatis) study for the Greek economy using i/o data
for the year 1970 finds a MAD in the range of 16-20% and in this range
the rate of profit and the rate of surplus value estimated either in
values or prices of production do not differ in any significant way from
those estimated in market prices. In this study we also estimate the
correlation coefficients which in no case are 99% but rather in the
vicinity of 90% (87% to 92%) and they are consistent with the other
measures of deviation. A result (of consistency of the correlation
coefficient and the other summary statistics) which is also found in the
studies of the US economy.Thus the correlation coefficient can be used
in conjunction with the other summary statistics as an additional
measure of deviation.

One of the most difficult issues that we faced in our study was the
quality of data and the quality of data gets worse in the case of
employment statistics, which are the most crucial statistics for the
verification of the labor theory of value. This is not an issue peculiar
to the Greek economy but rather it is general and might include even
the US economy. For example, how is one to account for the self employed
population and the capitalists engaged in each particular industry? And
then how does one reduce this labor to simple (abstract) labor time. In
such a case to be precise you need labor hours and the data are hard to
come by for all economies.

There is also the question of productive unproductive labor which needs
to be accounted for in the estimation of values.

The other question is the treatment of depreciation, where there are a
lot of arbitrary assumptions, nevertheless depreciation one way or
another must be accounted for.

The question of capacity utilization must also be addressed for each
industry of the analysis.

Under these limitations (and the list could expand) when one finds even
a 20 to 25 percent average deviation of values from market prices to my
view this demonstrates the strength of the theory which (despite the
lack of adequate data) nevertheless shows that the variations in market
prices are within well specified bounds, narrow as in the case of the US
and former Yugoslavia relative narrow as in the case of Italy, Greece
and England. In the Case of Mexico, however, it really is hard to
accept an 80 0eviation as verification of the LTV. I would be curious
to hear about the quality of the employment data of the studies on
Mexico and England.

In solidarity

Lefteris Tsoulfidis