Re: [OPE-L] measuring causes[MESSAGE NOT SCANNED]

From: Paul Cockshott (wpc@DCS.GLA.AC.UK)
Date: Fri Dec 23 2005 - 17:23:25 EST


It partly depends on whether a equational relation between
economic variables can be interpreted as a directional causal
relation.
For example the equation p'= s/(c+v)
relates 4 economic variables.
Suppose the rate of profit has fallen and the organic
composition has risen.
Just from the equations we can as easily decide that
the rise in organic composition has caused the fall in
profit or that the fall in the rate of profit had
caused a rise in the organic composition. Either
is compatible with the equations which merely specify
mutual constraints within the system.

This specification of mutual constraints is true
of any system subject to a mechanism or set of laws
of motion.

But as economists we would be loath to say it was
the rate of profit that caused the rise in the organic
composition of capital.

We would treat organic composition as causal and 
the rate of profit as an effect rather than vice versa.

I would disagree with allin when he says that to 
identify separate causal processes the components - 
s/v  c/v etc have to be mutually orthogonal. 

I think this is too strong, for them to be separate
causal processes one only needs the components to
be linearly independent which is a weaker condition
than orthogonality.



-----Original Message-----
From: OPE-L [mailto:OPE-L@SUS.CSUCHICO.EDU] On Behalf Of Allin Cottrell
Sent: 21 December 2005 02:09
To: OPE-L@SUS.CSUCHICO.EDU
Subject: Re: [OPE-L] measuring causes[MESSAGE NOT SCANNED]

On Sat, 17 Dec 2005, michael a. lebowitz wrote:

> Paul,
>        What you are doing is decomposing the rate of profit (and
> there could be other variations-- eg., taking into account turnover
> insofar as the appropriate rate of surplus value is the annual one),
> but is that the same as identifying causes?

I for one will answer 'No'; I think Michael has reasonable grounds
for skepticism.

When we have a quantifiable variable in view -- for example, the
rate of profit, or per capita income, or a country's imports -- it
is generally possible to produce an "accounting decomposition" of
any change in the variable in question.  Here are two simple
examples:

1) we can decompose a change in per capita income into a change in
aggregate income and a change in population.

2) We can decompose a country's imports into a change in the
imports-to-GDP ratio and a change in GDP.

The examples given show that the quantifiable variable in question
need not be inherently "compound" or "complex".  In the first case,
per capita income seems inherently a complex concept (a ratio); but
in the second, the volume of imports seems "simple".  An "accounting
decomposition" is possible just on condition that the variable in
question can be expressed as the product or quotient (or other more
complex mathematical resultant of) two or more other quantifiable
variables.

Consider per capita income.  Suppose that in some country over some
period, per capita income falls by z percent.  Modulo some finagling
of percentage changes, we can say, in the accounting sense, that
this is due to the extent (x/(x+y)) to an x percent rise in
population, and to the extent (y/(x+y)) to a fall in aggregate
income of y percent.

But wait -- if an economy is "doing OK" we don't expect a rise in
population to be automatically accompanied by a fall in per capita
income.  Rather, we expect aggregate income to rise in line with
population, other things equal.  So we resist the idea that a rise
in population "explains" a fall in per capita income, in a true
causal sense as opposed to a merely arithmetical sense.  We want to
know why aggregate income failed to rise in line with population
in this case.

Let's return to the rate of profit.  As Paul has noted, movements in
the rate of profit can be decomposed, in an accounting sense, into
movements in (a) the rate of surplus value, (b) the organic
composition of capital, and (c) the proportion of total surplus
value appearing as profit.

Here there are stronger grounds for mapping from an accounting
decomposition to a causal decomposition, since the components (a) to
(c) relate directly to definite theories proposed by various Marxist
writers to account for historical movements in the rate of profit
(as opposed to their being merely arbitrary mathematical
"components" of the profit rate).

Nonetheless, there remain some grounds for skepticism. The
"decomposition" approach implicitly assumes that the identified
components are mutually orthogonal -- that is, that it makes good
sense to consider the counter-factual, "What if A had changed but B
had not?"

For example: A rise in organic composition was responsible for the
fall in the profit rate to the extent of x percent, since if the
organic composition had not risen, then -- "other things equal" --
the fall in the rate of profit would have been x percent less.

But Marx, for one, linked changes in organic composition and the
rate of surplus value, considering them to be joint effects of
changes in production methods that substitute fixed capital for
labour.  (Use machinery in place of living labour: org comp goes up,
and so does s/v.)  From this point of view, the orthogonality
assumption does not hold, and therefore the accounting decomposition
does not map directly onto a causal decomposition.

Allin Cottrell


This archive was generated by hypermail 2.1.5 : Sun Dec 25 2005 - 00:00:02 EST