[OPE-L:8121] Re: Re: Marx and the labour theory of value

From: clyder@gn.apc.org
Date: Wed Dec 04 2002 - 17:09:22 EST

Quoting Michael Eldred <artefact@t-online.de>:

> > > An alternative procedure converts the wage bill of each industry to
> > a quantity of hours by dividing through by the average hourly wage
> > in that industry. One then obtains a row in the i/o matrix that is
> > in terms of hours. One can then apply the recursive adding up
> > method that Marx uses in Capital to get the labour content in hours
> > of the outputs of all industries, this gives us a value vector in hours
> which
> > you can compare with the vector of final selling prices.
> Only the second method is at all relevant for the LTV and it assumes that the
> total
> (recursively added up) wage bill, which can be broken down into the total
> number of
> hours of labour performed, has something to do with the magnitude of value
> considered
> as the socially necessary labour time embodied in the commodity product.
> Access to
> labour-times is only _through_ the wage value-form, which is used to obtain a
> number of
> hours. But how does one distinguish between the various purported
> value-creating
> potentials of different concrete labours? 

I dont see that this is relevant at this point, though there are
other issues for which it is relevant.

The method of deriving hours worked by dividing by the mean hourly rate
in each industry has the effect of reducing all labour to hours abstracting
a) the different concrete forms of labour in different industries
b) the possible different skill levels in different industries reflected in
   some industries having higher wage levels than others.

This then acts as a form of sensitivity analysis - if we get a 95% correlation
even with this forced reduction of all labours to simple labour, then the
effects of differential value creating powers of different labours must
be quite modest. That is not to say it does not exist, just that it is not
an enormously influential effect.

> > PC: The R^2 that one obtains from the latter method for the UK is about
> 0.95 as
> > against
> > about 0.97 for the former method.

I said that from memory, checking it again today the ratio was the other
way round, one got a slightly better result when expressing everything
in terms of actual hours than when one simply used vertically integrated
labour coefficients. This would indicate that inter-sectoral differences
in wage rates reflect more differences in the rate of exploitation than
differences in value creating power.

> > The regression coefficients are impressive (which I will accept for the sake
> of the
> argument, without studying the mathematics in detail --Paul Cockshott and
> Allin
> Cottrell "Does Marx Need To Transform?" http://www.wfu.edu/~cottrell/vol3.pdf
> ).
> Empirically applied probability distributions are able to extract
> regularities from
> masses of data previously not amenable to mathematical analysis. But, as far
> as I can
> see at the moment, the latter method too makes the same assumption regarding
> complex
> labour -- or else it reduces all labour to simple labour. The high
> correlation you
> obtain seems to be that between total recursive labour input costs and the
> selling
> prices of commodities. Since these total wage costs include the break-down of
> constant
> capital into wage costs, and rent components from natural conditions of
> monopoly are
> excluded as "outliers", this is not surprising, since the sine qua non of
> capitalist
> production is that advanced capital returns at least without having been
> diminished.

> Employing the aggregative methods of mathematical statistics is like seeking
> shelter in
> large numbers -- in the long run and on the whole, capital manages to
> valorize, i.e. to
> more than recover its total recursive labour input costs.

If we just did that, your criticism would be justified, but we only
excluded 2 industries out of some 80 or so. We also publish the figures
for correlation with and without the oil industry and both are strong.

> >
> > ME: How does one compare "values and prices"?
> > --------------------------------------------------------------------
> > Paul replies:
> > There are a number of different metrics used by researchers, one can
> > correlate the two vectors, one can measure the angle between them,
> > one can compute their normalised inner product.
> >
> > The important thing to recognise is that one has a vector of prices
> > and a vector of values for the same set of commodities, one can then
> > apply any of a number of well known techniques to compare the similiarity
> > of these vectors.
> Once one has made the leap from the temporal to the monetary dimension, the
> entire
> apparatus of mathematical statistics stands at one's disposal. That is why I
> am
> concentrating on how times are derived from the data precisely through the
> value-form
> of wages.

What is wrong with using wages to work back to time ?
We had access to hours worked per person per week for each industry and
for weekly wages. We also had the total wage bill of each industry. Making
the assumption that there were 20 days holiday a year, one can from this
get a pretty good estimate of the number of hours worked in each industry.
> > ----------------------------------------------------------
> > ME: To take just a minor point: How does one
> > take into account factors such as fertility of the soil (for agricultural
> > products) or the differences between monetary currencies in different
> economies?
> > -------------------------------------------------------------
> > PC:
> >
> > This is a good question. Remember that we are dealing with
> > the aggregate national product of whole industries.
> > Differences of fertility thus do not affect the figures directly
> > but there is an indirect effect insofar as the profits of some
> > industries incorporate a significant portion of rent. In the UK
> > case this is most significant for the crude oil industry. One finds
> > that its market price is significantly higher than would be expected
> > on the basis of its embodied labour content. The same applies
> > to the petroleum refining industry. In these cases the deviation
> > of prices from embodied labour content is in line with what
> > one would expect from the Ricardian theory of differential rent.
> >
> > The differences in currencies between different economies are
> > dealt with as follows.
> >
> > There is a row vector in the table indicating the exports of each
> > industry, and a column vector indicating the imports. Were trade
> > in balance one could simply take the total domestic labour content of the
> > exports and divide it pro rata among the imports used by each
> > industry - since according to Smith, domestic labour is the real
> > cost of imports to a country. Since trade is not typically in balance
> > one evaluates the mean domestic labour content per Pound
> > Sterling exported in the UK case for British exports, and then
> > imputes this to each Pound Sterling spent by each industry
> > on its imports.
> >
> > > PC: It is my experience of doing empirical investigations that they
> almost
> > > always teach you something new that would not have occured to you had
> > > you not gone to the trouble of doing them.
> >
> > ME: My objection is that the preconceptions (i.e. one's concepts) determine
> in
> > advance
> > what is to be empirically measured.
> > -----------------------------------------------------------------------
> > PC: This is obviously the case, in that if one had no concepts one would
> not
> > know what one wants to measure. But one's preconceptions do not
> > determine the results of what you measure.
> That's true. But the results also have to be interpreted. What I am
> questioning is
> whether the empirical analysis you describe is a test of the LTV at all but
> instead
> represents a validation that capitalism is capitalism, i.e. that on the
> whole, money
> capital advanced returns augmented.

I think that in essence the law of value derives from this 
very point.

> ---------------------------------------------------------------------------
> > PC: I don't agree with this. The I/O table was an invention deriving from
> the
> > method of material balances adopted by Gosplan. This in turn derived from
> > the real need that any socialist economy has to determine the socially
> necessary
> > labour time required for projects. The i/o tables prepared by western
> > states are deficient in that they contain limited data on disaggregated
> labour
> > types used, but they already encapsulate the aspect of social necessity by
> > considering the labour of the entire society at once, much as Engels
> hypothesised
> > would be done in a socialist economy.
> The fact that I/O tables were invented in the context of USSR production
> planning
> through Gosplan, I think, highlights my objection. In a total social
> production plan,
> production inputs, including labour inputs, are sociated in the plan itself
> -- and not
> via the market. The sociated form of the products of labour is not (monetary)
> value but
> the overall plan itself in which each kind of labour and each product of
> labour has a
> concrete, differentiated, albeit aggregated position -- not simply an
> abstract validity
> as an amount of money.
> Gosplan represents an historical attempt at consciously sociated labour. The
> conscious
> planning by some sort of political apparatus replaces the sociation via the
> marketplace. The markets become merely places of distribution according to
> foreseeing,
> precalculative planning. Instead of the dissociated self-interests of
> producers,
> workers, managers, landowners, financiers, etc. primarily driving economic
> activity
> through the mediation of the various sociating value-forms, the motor of
> consciously
> sociated socialist production is politics itself, with all the inertial
> inefficiencies
> of bureaucracy and all the conscious political manipulation and individual
> abuse that
> this is heir to.

I agree with this point in general.

The i/o tables we have available are deficient compared to those available
in planned economies, and one has to go to considerable pains to extract the
real data from the monetary form in which they are presented. Allin 
has demonstrated however that there are valid procedures by which this can
be done. If I recall he did this in a paper to the Working Group on Value
theory a few years ago.


This archive was generated by hypermail 2.1.5 : Fri Dec 06 2002 - 00:00:00 EST