[OPE-L:3781] Re: Re: On the Unimportance of the Transformation Problem

From: Andrew_Kliman (Andrew_Kliman@email.msn.com)
Date: Thu Sep 07 2000 - 09:57:59 EDT

[ show plain text ]

Duncan commented that "Wasn't it Shaikh who first referred to a 93%
prices of production as reflective of values? Why not just take it as
100% and move on to other issues?"

Paul Cockshott replied, in OPE-L 3776, "I would side with you here. All
the empirical studies that I know of apart from Steedmans back this up,
many even more strongly."

Well, Paul surely knows about my own empirical study, which is
forthcoming in the CJE. It finds that, once one resolves the problem of
spurious correlation by controlling for the variations in industry size
that produce spurious correlation, there is *no* reliable evidence that
sectoral values have any influence upon sectoral prices. The
cross-sectional correlations, which are often much larger than 0.93 (in
the case of my study, about 0.99) when researchers fail to control for
industry size, become very small, insignificiantly different from zero,
and quite often negative once the spurious correlation is eliminated.

I did not examine the relation between production prices and values, but
since prior studies concerning this relation have also been marred by the
same spurious correlation, the reported correlations are, in light of my
findings, highly suspect.

My paper also finds that the "closeness" of values and prices does not
mean what people think it means. It is NOT the case that a sector's
(size-adjusted) price is particularly close to its OWN (size-adjusted)
value. Prices and values are "close" only in the sense that all the data
are tightly bunched together -- almost all sectors' prices are close to
almost all sectors' values.

Andrew Kliman

This archive was generated by hypermail 2b29 : Sat Sep 30 2000 - 00:00:04 EDT