Re: [OPE-L] why does the debate on the "transformation problem" continue?

From: Paul Cockshott (wpc@DCS.GLA.AC.UK)
Date: Thu Mar 29 2007 - 04:26:08 EDT

> The difference is that if you have a non uniform rate of profit
> 1. You can choose the case where it is a null transform - which is
> certainly no less unlikely,
>     than the uniform profit rate.
I think this is my point. In such a case how can the empirical
demonstration that market prices are close to standard labour-values
distinguish between
- confirmation of the predictive power of standard labour-values
- confirmation that the distribution of profit rates is negatively
correlated with organic compositions?
In the latter case, there is no need to transform.  


When I said above "You can choose the case where it is a null transform"
I did not mean that you can choose that independent of the empirical data.
I was saying choose in the sense that you were 'choosing' to study the
hypothetical case of a uniform rate of profit. One could similarly 
hypothetically choose to study the null transform.

But that is when one is doing speculative political economy. When one
is doing empirical political economy there is no 'choice', you can look
at the I/O table and see if the transform actually is null, or if it
is not. 

 if a conjectural event occurred in which profits moved closer to
uniformity then during that time period would you recommend we reject
the labour theory of value?


A single conjucture in which this occured would not be an adequate basis for
changing one's theory, though it should certainly give you pause for thought.
If one got a whole series of repeatable results from different countries and
time periods then one would definitely have to reconsider. One would have
to look at things like the actual deviations between total transformed
profit and total surplus value under normalisation conditions in which
total profit equaled total surplus value to assess whether the empirical
errors arising from calculating the rate of exploitation using the simple
labour theory of value were significant.

The latter point was the key argument of Steedman - that  one could say
nothing about the rate of profit from the rate of surplus value and
organic composition calculated in value terms. However, a mere equalisation
of the rate of profit would not be enough for Steedmans objections to
have empirical weight. It is highly probable that because of the random
mix of constant capital goods, which themselves would contain a spectrum
of PP/value ratios, that the errors on the aggregates would still be small.

If on the other hand one found that real economies had the kind of i/o table
structure that Steedman uses in his book, and if profit equalisation did
occur as he assumes, then his objections to the labour theory of value
would have a lot more force.

But these are empirical questions, and as scientists we have to adapt our
theories to what actually exists. The current data does not seem to support
Steedman's objections at all.

> 2. If it is not either the null transform or the uniform profit rate you
> are providing a whole
>     lot of new variables which can be set to whatever you actually
> observe the profits and
>     market prices to be. At this point the theory no longer has any
> predictive ability ( there
>     is no economy of information in comparing the theory with the actual
> observed data).
>    Thus it is no longer a theory, merely a record of observed data.

The TP is not about whether labour-values are good predictors for
market prices. The TP is whether, under the assumption of a uniform
rate of profit, labour-value is conserved in prices of production.


I think you are engaged in a partial presentation here. One issue
that has been raised is to do with the conservation of the constraints
of total value = total price and total sv = total profit, yes, 
but that is only part of the question.

The issue only arises because it is hypothesised that one has
to modify ones theory of price to take into account an assumed
real world equalisation of the rate of profit. The question you
are addressing is a derived question that only arises in the context
of a theory of price. The prior question is how good is the theory
of price.

A price theory must attempt to predict exchange values on the basis
of something other than the observed exchange values themselves. The
classical labour theory of value and Sraffian prices of production both
do that. The ltv does it on the basis of the technology matrix and 
labour input vector, price of production theory requires that plus 
additional data on the distribution of aggregate data between classes.

PP theory is thus of weaker predictive ability than LTV since one needs
an additional variable to be supplied to the model to produce an answer,
adding this extra variable would only be justified if the resulting
estimate of prices improved sufficiently to compensate for the additional
information one was putting into the model by supplying the class distribution
of income.

If one proposes models in which the actual sectoral profit rates are
taken as input variables, as you did in your previous email, then 
one has a very weak theory indeed, as you have added as many new 
variables as the prices you are attempting predict. Such a TP theory
would be of no scientific interest.

I will reply to other questions later when I have time
Saying that in reality there is not a uniform rate of profit doesn't
get to the root of the issue. That response does not answer the
theoretical problem that a uniform rate of profit appears to prevent
labour-value conservation. Why should this be so?

Marx proposed the transformation in order to maintain conservation
while following the classicals in maintaining that individual prices
diverge from labour-values due to a general rate of profit.

What is your answer to this theoretical question? - That such a state
cannot occur in reality?, or that if it did occur then conservation
will not hold?

> A scientific theory must be more concise than the data it purports to
> explain, if you introduce
> as many free variables as the market prices, it becomes vacuous.

I cannot disagree. But as you know simplicity has to be traded off
with predictive accuracy. Otherwise, Newton should have not introduced
elliptical orbits, since circular orbits explained almost all the
data, apart from some small error. I'd accept a few more variables if
they explain all the data.

Best wishes,

This archive was generated by hypermail 2.1.5 : Sat Mar 31 2007 - 01:00:12 EDT