[OPE-L:5624] ope-l: RE: In defence of correlation

andrew kliman (Andrew_Kliman@classic.msn.com)
Tue, 21 Oct 97 17:51:21 UT

[ show plain text ]

A response to the PIAF:

From: owner-ope-l@galaxy.csuchico.edu on behalf of Paul Cockshott
Sent: Monday, October 20, 1997 4:27 PM
To: andrew kliman; ope-l@galaxy.csuchico.edu
Subject: [OPE-L:5624] Re: ope-l: RE: In defence of correlation

Paul: "All of the examples they andrew uses apart from energy usage are in
effect value or price measures. If values and prices are correlated, then
these will also tend to be correlated as a side effect."

I expect that any quasi-physical measure of "capital" stock, total assets,
advertising expenditures, as well as energy usage, will be positively
correlated with aggregate sectoral price as well. If that's the case, then a
positive correlation between aggregate sectoral value and aggregate sectoral
price has no special significance -- almost all measures of industry size will
be positively correlated.

Paul: "If one were to be take a reasonable null hypothesis for something that
would be expected to have a low correlation one would have to chose something
like the net weight of the industry's output against its price."

The point is not to find something with a low correlation, but to point out
that almost all measures of industry size will in fact be positively
correlated. I think even net weight would be correlated with other measures
of industry size, such as aggregate sectoral price, except that net weight is
a meaningless measure when it comes to transportation, utilities, and

I had written: "If [the C-C theory] is unable to generate hypotheses that
allow one to discriminate between it and the naive hypothesis, then the two
are identical for practical purposes, and C-C's claims reduce to the obvious
point that firms recoup their costs and get some random cut of surplus-value."

Paul responds: "In a sense that is what the Farjoun and Machover theory of
value/price amounts to. What is remarkable is what powerful predictions they
are able to make from that basic proposition given a more sophisticated
definition of what a 'random' cut of surplus-value means."

I'll have to check out F-M more closely. In any case, I'm not surprised that
the naive hypothesis -- firms recoup their costs and get a random cut of
surplus-value -- generates powerful predictions. Moreover, if your claims do
in fact reduce to the naive hypothesis, I think it would help readers
interpret your work and the evidence if you were to emphasize that point. I
think much of the work in this area, and particularly the labeling of the
naive hypothesis as a "labor theory of value" gives people the impression that
values are actually much closer to prices than would be expected were
surplus-value randomly distributed.

I had written that the claim that "values" are good predictors of prices is
generally understood to imply, in part, that they are "unbiased" predictors of
prices: a sector's mean price will equal its value. The evidence we have
indicates that this isn't true."

Paul replies: "This is wrong. Values are unbiased predictors of prices in
this sense. The price value ratio, taken over all industries has a mean of 1."

Please! What's at issue is not the *total economy's* mean price/value ratio,
but the relationship between a *particular sector's* price and value. A
sector's price/value ratio will differ across different observations of that
sector. A lack of bias means that the observed price/value ratios in that
*particular* sector have a mean of 1.

Assume we have two sectors of equal size, and 1000 observations on each.
Assume that the price/value ratio in the first sector is always 1.99, and the
price/value ratio in the second is always 0.01. Then the total economy's mean
ratio is 1, but each sector's price is very far from its value, and values are
a horrible predictor, very biased. If the total value in each sector is 100,
then the C-C theory predicts that total price will average 100 in each sector.
But in the first sector it averages 199, and in the second sector it averages

Since Paul is an empiricist, and since his claim that values are good
predictors of prices is supposedly an empirical one, let me also ask what
possible evidence could falsify the claim that the weighted average of
price/value ratios across the economy is 1?

I had noted: "Enough evidence -- some coming from C-C themselves -- has been
reported on this list to allow us to conclude that prices deviate
systematically from values due to rent but also due to the fact that sectors
with larger capital/labor ratios tend to have higher price/value ratios. "

Paul replied: "This is true, but that does not make values a biased price
predictor. To the extent that oil and oil related products sell above their
values, other commodities sell below their values to compensate, so the
predictor remains unbiased."

I think readers deserve to know that sectoral prices deviate systematically
from values, the direction of that deviation, the size of the deviation, and
the importance of it for the hypotheses in question.

Again, I do not think that Paul's notion of "unbiased" is a meaningful one.

I had written: "Paul Cockshott, for instance, reported that prices on average
lie midway between values and production prices. So values are a biased
predictor of market prices."

Paul replies:
"This is not a question of bias ...."

For the reason I've stated above, I reiterate that it IS a question of bias.

Paul: "These results have very considerable implications.

"On the ideological level they cut the ground from under those economist
like Steadman who have argued that the labour theory of value is unnecessary
in the Occam sense, ...

"This has great propaganda implications. If one can prove simply and
directly that labour creates value, then it is much easier to
expose exploitation than if one adopts the price of production theory,
which, at first sight, implies that capital creates value."

If one thinks that the strength of the naive hypothesis has considerable
ideological and propaganda implications, so be it. I suspect that the
implications come rather from the propagandistic manner of discussing the
hypothesis and the evidence.

I do not think the non-falsification of the naive hypothesis "proves" any
theory of value. As I have noted, I strongly doubt that ANY economist, dead
or alive, has ever disputed that prices will be reasonable close to costs plus
a random share of total profit.

Paul: "Andrew will have to be a bit more specific as to what he means by a
random distribution of surplus value before we can answer this question."

Well, I think Allin has understood what I've said about this, and has
responded to it. I hope to address his post soon.

Andrew Kliman