[OPE-L:2539] Re: Marxian empirical research

Paul Cockshott (wpc@cs.strath.ac.uk)
Wed, 19 Jun 1996 02:39:50 -0700 (PDT)

[ show plain text ]

>I have been very interested in the recent discussion between Duncan and Paul
>C. about Marxian empirical research, and the justification for doing Marxian
>empirical research in terms of money and prices, rather than in terms of
>Paul explained that he and Allin were originally motivated to do their
>empirical work on the correlation between prices and labor-times to overcome
>objections by referees from Capital and Class that their estimates of the
>rate of profit and the rate of exploitation were in terms of prices, and
>that this was illegitimate because Marx's concepts are in terms of
>labor-values. Their estimates showing a strong correlation between
>individual prices and values provided a justification for their macro
>estimates in terms of prices.

Paul C:
There has been one other motivation for our work, we are in addition
concerned to justify our advocacy of what we take to be the classical
socialist position that economic calculation in a socialist commonwealth
would be in terms of labour time, against the arguments of those like
Roemer who argue that a socialist economy must rely upon prices and
the market.

This comes up in particular in the context of the next point that Fred

>I recognize of course that according to Marx's theory, these observable
>phenomena of money and prices are explained by quantities of labor-time.
>But the labor that explains the quantities of money and prices is ABSTRACT
>SOCIAL LABOR, which is not directly observable as such. The unit of
>abstract labor is labor without special skills and of average intensity. No
>one has yet (so far as I know) devised an appropriate way to estimate
>abstract labor. This requires a conversion of skilled labor to unskilled
>labor and a conversion of labor of unequal intensities to labor of average
>intensity. Therefore, it appears to be very difficult, and perhaps
>impossible, to derive rigorous estimates of abstract labor. Not only is it
>not necessary to justify Marxian empirical research in terms of prices by
>means of correlation between prices and labor-times, it may even be
>impossible to do so rigorously (i.e. in a way that is consistent with Marx's
>concept of abstract labor).
>It is not clear to me how Paul and Allin estimate abstract labor in their
>Capital and Capital article. They mention that they tried two different
>methods to estimate abstract labor, and refer to "discussions of Models A
>and C below" for further description of these methods, but I was not able to
>find these discussions in the article. Perhaps the Capital and Class
>referees, after insisting that you do this unnecessary empirical test, then
>made you cut the description of your estimation procedures.
>Paul or Allin: would you please send me this description of your methods of

Paul C:
We take the position that the problems that one has in understanding a
lower mode of production can be solved if you consider the more advanced form
that suceeds it. Thus to understand value and abstract labour properly you
must as Marx often did, consider the matter from the standpoint of a future
communistic society.

Our initial analysis of the problem of how to define abstract labour is thus
from the context of how a socialist planned economy could perform its
calculation in terms of labour. One can then look back at capitalism and
understand to what extent the same things appear there in fetishitic form.

Abstract/Concrete and Skilled/Unskilled
If one is to perform calculations in terms of labour time, one must
both abstract from the concrete character of labour - the branch of
production and trade in which it is exercised and take into account
individual differences in proficiency.

These are two quite distinct issues, and it is important that one
does not confuse them. The use of the word 'skill' is not helpfull
here as it has two meanings - the difference between the skills
of an electronic technician and traffic supervisor being one meaning.
Another meaning refers to the difference between a profficient
and less proficient individual within one trade - between a good
electronic technician and a poor one. Let us call the first
sense of skill concrete skill and the second sense, level of

In the short to medium run, the differentiation of labour by concrete skill
is both
important and irreducible. The skills of a mining engineer, a surgeon and a
computer programmer are not interchangeable. It follows that over this time
horizon planners cannot simply think in terms of the allocation of `labour'
as such, but must recognise the constraints imposed by the availability of
specific skills. This implies that detailed records should be kept of the
number of people qualified in each speciality. But then what becomes of the
labour conception of value and the use of abstract social labour as a unit of

Well, in the long run workers can be retrained, and the `democratic'
assumption of socialists is that, apart from certain extremely demanding tasks
and certain impaired individuals, almost everyone can do almost everything.
In the context of long run planning, what matters is not the present
of specific types of skilled labour, but rather the cost of production of those
skills. And just as the value of machines can be calculated in terms of the
amount of labour time required to make them, for the purposes of long term
economic calculation, so can human skills. Long run economic planning
would involve plans about resources to be allocated to education and
training. The assumption is that if changes in technology imply that the
economy will need 10,000 people trained in cabling optical fibres, then,
given the appropriate training program 10,000 people who in a previous
generation might moved into some other trades can be concreteised as
laser optics technicians.

In this sense, abstract labour is the undifferentiated productive
potential of the young human animal, which society can concretise in different
forms through its educational and technical formation system.
This is the fundamental resource that is available to human society,
the polymorphous productive potential that must make it the starting
point of economic planning.

It is abstract in the sense that, with the same population in 10 years time,
there is an imensity of different concrete ways in which that population
could be employed. There remains the constrain, however, that if society choses
to develop certain branches of production that require a longer training
period, then it must allocate additional time to forming this labour
force. It must add in the time spent by the trainees themselves and
by those that train them, since both of these quanta of time are drawn
ultimately from society's common pool. How could one calculate this?

The answer that we gave in our book is as follows:

Illustrative calculation of skilled labour multiplier

This explains in more detail the calculation of the skilled labour
We first illustrate the calculation of the total embodied labour
content of skilled labour.

1. On the part of the student. Assume 4 years of study at 40 hours per week
for 45
weeks per year.
Total: 7200 hours.

2. Classroom teaching. Assume 15 hours per week, 35 weeks per year, for 4
distributed across an average class size of 30 (average of large lecture classes
and smaller labs, seminars etc.).
Total per student: 70 hours.

3. Tutorial work. Assume 2 hours per week, 30 weeks per year of one-on-one
Over 4 years, total = 240 hours.

4. Educational overheads. Let us suppose this amounts to a contribution
equal to
the classroom teaching labour.
Total 70 hours.

Now examine the breakdown of this total labour content into simple and
skilled. The
student's own contribution is simple; the teachers' contribution is skilled;
and let us
assume for the sake of argument that the `overhead' contribution breaks down
50/50 skilled and unskilled. We then arrive at the following: total labour
content of
skill production equals approximately 7,600 hours (rounding up), of which
labour makes up around 5 per cent (rounding up again).

The total embodied hours figure quoted above is a first approximation (in
fact an
underestimate, as we shall see). Let us denote this approximation by TH(0).
TH(0) we can construct a first approximation to the transmission rate of
labour on the part of skilled labour:
R(0) = TH(0)/AH.D
where AH represents the annual hours the skilled worker will work once
and D is the depreciation horizon in years. We can now use R(0) to
re-evaluate the
total hours embodied (on the assumption that the transmission rate for the
and others who supply the skilled input into the production of skilled
labour is the
same as that for their students, once qualified). If the proportion of
TH(0) accounted
for by skilled labour input is denoted by SP, our revised estimate of the total
embodied labour is
(1+R(0))SP.TH(0) + (1-SP)TH(0) = (1 + R(0)SP)TH(0).

But this new figure for total hours embodied can now be used to re-estimate the
transmission rate, permitting a further re-estimation of total hours-and so on,
recursively. The resulting successive approximations to the total labour
in the production of skilled labour form a geometric expansion, the nth term
of which

(1 + R(0)SP + R(0)^2 SP^2 + R(0)^3 SP^3 + ... + R(0)^n SP^n)TH(0).

Letting n tend to infinity, we can deduce the final limiting value of the
total hours
estimate, namely (1-R(0)SP)^-1 TH(0), and the corresponding final estimate
of the
transmission rate for embodied labour
Rf = (1-R(0)SP)^-1 TH(0)/AH.D.
Remembering that R(0) = TH(0)/AH.D, Rf may be rewritten as
Rf = TH(0)/(AH.D-SP.TH(0)),
enabling us to calculate the final transmission rate directly. Using the above
illustrative figures of TH(0) = 7600, AH = 1575 and SP = 0.05 we find that
Rf = 0.50 for D = 10,
Rf = 0.33 for D = 15,
Rf = 0.24 for D = 20,
as quoted in the text. In each case the skilled labour multiplier is simply
1 plus Rf.

Applied to empirical analysis of capitalist economies

When dealing with capitalist economies, for which the economic
statistics are much less detailed than they would be in a
socialist one, what one is presented with is the total wage
bills for different industries. This is not broken down into
wages on different trades without a great deal of extra
effort in collating statistics from different sources.

There is no problem in converting concrete labour to
abstract labour, since this operation of abstraction is
performed as soon as one treats labour in different industries
as simply 'labour'. What is a potential problem is the possibility
that some industries require a higher degree of mean training
for their work forces than others.

Our Method A and Method B attempted to deal with this by using
wage rates as a surrogate for training costs. The assumption is
that wage rates are positively correlated with the labour time
required to train sombody.

Method A took the total wage bill for an industry as a surrogate
for its labour input. Method B adjusted this to take into account
industry wage differentials. Thus an industry which paid low wages
would have, under Method B its wage bill divided by its wage rate,
and would in consequence be counted as employing more labour than
in Method A. Method B amounted to assuming that inter industry
differences in wage rates were adventitious and unrelated to training

We found that Method A yielded a closer correlation between prices
and labour content than method B.
Paul Cockshott