>
>
>Paul wrote:
>
> For any interpretation of Marx that you make to be consistent with
> the fact that only a small proportion of commodities are sold at a loss,
> you have to assume that the distribution of prices around values is
> reasonably tight.
>
Fred repplied:
>I do not see why this is true. Consider the following:
>
>1. Assume a wide distribution of compositions of capital across industries.
>
>2. Assume prices of production which equalize rates of profit (as in Marx's
>theory). These prices of production will not necessarily have a tight
>distribution around values (one can always choose compositions of capital
>that generate wider distributions of prices around values).
>
>3. However, since all capitals receive the general rate of profit, no
>commodities are sold at a loss.
>
>
>Therefore, it is not necessary to assume a close correlation between prices
>and values in order to be consistent with no commodities selling at a loss.
>
Paul Comes back:
I think that you and other people have been misled by considering innordinately
small models.
Try constructing a model with a couple of hundred industries and a roughly
normal
distribution of organic compositions of capital - say clustered around 4 with
outliers as high as 8 and as low as 0.5. Also take into account that 20 to
30 inputs go into making up the constant capital of each industry. Assume also
a realistic rate of profit say 10%. You will
find the following:
1. Because the inputs to each industry are a random sample from industries
with high and low organic compositions, you will find that the price and
values of the aggregate constant capitals used in different industries
are highly correlated. (This is the real reason why it is scarcely necessary
to transform inputs.)
2. Since no commodities sell at a loss, then given (1) we can say that very
few commodities will sell for less than c+v expressed in value terms.
3. You will find a profit/wage ratio ranging from a low of 15% to a high of
90%, but with most of it clustered in the range 35% to 65%.
4. The price/value ratio will range from 82% to 104%, but most of it will be
clustered around 96% to 102% as shown below.
c v s p val price price/val
0.5 1 0.5 0.15 2 1.65 0.825
2.5 1 0.5 0.35 4 3.85 0.9625
4 1 0.5 0.5 5.5 5.5 1
5.5 1 0.5 0.65 7 7.15 1.021429
8 1 0.5 0.9 9.5 9.9 1.042105
Where you go wrong is to say "(one can always choose compositions of capital
>that generate wider distributions of prices around values)". This is an
idealist conceit of the Sraffians. It is absolute nonsense. Economists are
definitely not at liberty to 'choose organic compositions of capital'.
That is something that is done in the real economy, all that we can do
is to try and understand the statistical properties of the distributions
of organic composition that actually arise. It can not be too strongly
emphasised that in constructing models of the real world one must observe
the statistical properties of the world. Any chaotic system tends to
move to a high entropy state, one is no longer modeling the world if
one selects low entropy distributions of ones statistics.
Paul Cockshott
wpc@cs.strath.ac.uk
http://www.cs.strath.ac.uk/CS/Biog/wpc/index.html