[OPE-L:3423] Re: Andrew's TSS and value added

Duncan K. Fole (dkf2@columbia.edu)
Tue, 15 Oct 1996 09:41:22 -0700 (PDT)

[ show plain text ]

In reply to Fred's [OPE-L:3399]:

>My response to Duncan is that the above logic of the determination of prices
>and the MVA is completely different from Andrew's own logic in his
>derivation of the falling rate of profit. Therefore, Duncan's critique is
>not valid.

I'm not sure that I have a "critique" of Andrew's method. I'm just trying
to clarify how it works.

Let me put the issue in a somewhat broader perspective. I find myself in
broad sympathy with the interpretation of the labor theory of value put
forward by the TSS group (if they don't mind being classed together for
this purpose): I think Marx's theory of value is a monetary theory, and
that it rests on the idea that money represents labor time and therefore on
the concept of a "monetary expression of value" a coefficient that
translates measures of social labor time into money. I also find myself in
broad sympathy with their methodological critique of stationary price
models, and their insistence that methods of analysis be robust enough to
handle nonstationary situations.

Thus I was surprised when Andrew produced examples that show the rate of
profit falling to zero under conditions of constant capital/output ratio
and steadily rising labor productivity in the 1-sector circulating capital
model, on the hypothesis that the monetary expression of value remains
constant. I was surprised because I thought I knew the answer to this
question, under precisely these hypotheses. My answer (which I've worked
out in posts to John and Andrew) is that the monetary rate of profit on
historical costs under these circumstances would be lower than the
commodity rate of profit because of the value losses on the inventories of
inputs during the period of production (as John emphasized in his
discussion of the issue), but would be constant, not falling to zero.
Furthermore, in my way of thinking about this problem, the price of output
in terms of money would fall steadily at the same rate as the increase in
labor productivity, rather than undergoing an accelerating fall as in
Andrew's examples. This difference is important, as Alan pointed out in his
eloquent posting on the problem of using the moving steady state solution
of a dynamical system to approximate its actual paths when these diverge.
In Andrew's examples they diverge, and in my understanding they don't.

Since we are doing the same problem and getting qualitatively different
answers, I wanted to understand the source of the difference. The source
turns out to be a difference in the definition (not the determination) of
the monetary expression of value. I would define the monetary expression of
value as the ratio of the traditional money value added to the social labor
time expended, which in the 1-sector circulating capital model leads to the

p(t)X-p(t)aX = m(t)l(t)X

I noticed that this is different from the starting point of Andrew's examples:

p(t)X-p(t-1)aX = p(t)X - p(t)aX +(p(t)aX - p(t-1)aX) = m(t)l(t)X

The difference is that in Andrew's examples the term (p(t)aX - p(t-1)aX),
which corresponds to the "inventory valuation adjustment" in National
Income Accounting jargon, appears on the lefthand side, whereas it doesn't
in my way of thinking about the situation.

Incidentally, this difference would not much difference if Andrew and I
were discussing the measurement of the monetary expression of value from
data on prices and output generated by a real economy. The IVA in most
cases is small, so the numbers we would get for m(t) would be quite

But, as you point out, Andrew makes a different use of these equations to
generate his examples. He makes the assumption that m(t) is constant (and,
without loss of generality equal to 1), and then uses the equation to
derive the path of prices. I have no problem with this general method or
"logic" as you put it. If you define the monetary expression of value by
one or the other of these two equations, it is logical to require price
paths that represent a constant monetary expression of value in an example
to satisfy whichever equation you think represents the correct definition
of the monetary expression of value. In fact, when I adopt Andrew's
equation and work out the price and profit rate paths on the basis of it, I
get the same answers he does. On the other hand when you solve the first
equation for the same assumptions you get the price and profit rate path I
described. Another way of putting the difference (which I used in one
response to John) is that the price paths in Andrew's examples do not
correspond to a constant monetary expression of value on my definition.

Although this difference in definition doesn't make much difference to the
measurement of the monetary expression of value given price and quantity
data in most practical situations, it makes a qualitative difference in the
price and profit rate paths in the examples. Without for the moment trying
to resolve the issue of which definition of the monetary expression of
value is appropriate to represent Marx or to give a price theory consistent
with actual experience, I think we can agree that the difference in the
results stems from the difference in these two equations. I read Andrew's
posts responding to me as acknowledging this point and moving on the
discussion of which of these two approaches it makes more sense to adopt.

(Actually, from Andrew's recent posts I see that the matter is a bit more
complicated than I have drawn it here, because from his point of view he
wants to include m(t-1), the monetary expression of value in the last
period, in the equation determining m(t). I can see the logic of this for
the TSS position, and that is why I asked Andrew to explain how he would
measure the monetary expression of value from arbitrary data on prices,
quantities, and social labor time expended in a 1-sector circulating
capital model.)

I do not think I am misrepresenting Andrew's method or logic.

>Duncan asks at the end of (3322):
> Do you think a capitalist economy with a constant positive rate of
> productivity increase and constant ratio of output to capital would
> experience a fall in the money rate of profit to zero?
>My answer: it depends on WHICH definition of the rate of profit one is
>talking about. If the rate of profit is evaluated at historical costs and
>the value of money is held constant, then under Andrew's further
>assumptions, the rate of profit would fall as in Andrew's example.

I confess I'm somewhat astounded that you should say this, since you have
done a lot of work on the movement of actual historical rates of profit,
which I have learned a lot from, and I never saw any sign of the type of
movement present in the TSS examples.


Duncan K. Foley
Department of Economics
Barnard College
New York, NY 10027
fax: (212)-854-8947
e-mail: dkf2@columbia.edu