[OPE-L:3307] Re: TSS and Value Added

andrew kliman (Andrew_Kliman@msn.com)
Mon, 7 Oct 1996 06:56:01 -0700 (PDT)

[ show plain text ]

I'm very busy now, and for the next several days, but I want to squeeze in
time to respond to Duncan's ope-l 3287. Plus one brief comment on his ope-l
3288, though a lot of the following refers to it implicitly. So here goes.

I'm glad Duncan laid out his underlying assumptions, because it helped me
realize why we're having difficulty communicating. There are some issues,
more elementary than those we've been discussing, about which we don't see
eye-to-eye, and which need to be clarified.

Duncan: "I think Andrew and I both start from an interpretation of the labor
theory of value that ...stipulates that in a given period an hour of expended
social labor creates a certain amount of money value."

I would usually have no reluctance in affirming this, but in the present
context, I'm wary of saying that expended living labor creates money value.
Even in the *aggregate*, money prices of commodities can rise or fall *after*
production for a myriad of reasons that have nothing to do with the living
labor expended *earlier*, *during* production. If, for instance, bank lending
increases markedly, money prices might rise such that equal portions of the
output, *produced* in a "given period" but sold in succession, may have
different money prices. Also, production takes time, so the relation between
labor-time and money value may vary during a given period of production.

Duncan: "Suppose we start with some data from a real economy that includes a
series of prices of output p(t), and a measurement of the quantity of output
X(t), and of the inputs to production, including a measure of the social labor
expended per unit of output l(t) (assuming that we've agreed on how to create
these measures.) Then I think Andrew and I would agree that the monetary
expression of value is equal to the ratio of the Money Value Added (defined in
the appropriate way) to the social labor time expended.

(1) MVA(t) = m(t)l(t)X(t)."

I don't agree with this, because production takes time. Only if production is
instantaneous do the l(t) and X(t) refer to one, and the same, t. In many
contexts, this issue, raised by Simon, need not present a problem, but in the
current context it does. Assume, for instance, that production is flow
input/point output. During one production "period," then, the living labor is
expended at a succession of times, all or almost all of them earlier than the
time of output. Consequently, here, and whenever production isn't
instantaneous, we can't assume a single uniform m(t) throughout the production

Duncan: "if we make the assumption that the monetary expression of value is
given (for example by production costs of a money commodity like gold), then
this equation turns into a theory of money prices, given the path of social
labor time expended. I think both Andrew and I agree that this is a meaningful
way to proceed."

Yes, *if* the givenness of the MEV means that it is assumed to be *constant*.
However, I'm unwilling to call this a "theory" of money prices, since it
follows from an arbitrary *assumption*, not from any claims concerning

Duncan: "The point I want to focus discussion around is that the price path
derived from this equation for any given path of social labor time is
sensitive to the exact definition of money value added we put on the left hand
side. Since both
Fred and Andrew seem to have misunderstood me on this point, let me emphasize
that I am not proposing any different treatment of the right hand side of this
equation (the measure of the social labor time)."

The measure of living labor-time is not my problem. Rather, if the MEV is
changing during the period, then *however* one measures MVA, I don't accept
the equation, because it assumes the MEV doesn't change.

Duncan: "Andrew's examples are based on the definition of the money value
added as the difference between the sales price of the output and the cost of
inputs which is (remembering that we are treating the cost of labor-power as

(3) MVA(Andrew) = p(t)X(t) - p(t-1)aX(t)

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

= MVA(National Income) + Inventory Valuation Adjustment

"In the interests of making the assumptions involved in an argument as
explicit as possible (as Alan advocated so eloquently in his discussion of the
Okishio literature), I want to call people's attention to the fact that this
definition of value added is different from the standard national income
accounting definition, in that it includes the revaluation of the stocks of
inputs over the production period due to the change in prices, which in NIA
terms is called the Inventory Valuation Adjustment, and is excluded from the
NIA definition of value added."

I'll accept all this, though I remain confused about the NIA definition, and I
still don't like the term "money value added."

Duncan: "I think it is also significant that by adopting this definition of
money value added to define the monetary expression of value one is
effectively imputing to the
expenditure of social labor the change in the value of stocks of inputs
through the production period due to price changes, a change in value which
has nothing to do with the production process itself. I confess that at this
point I think the NIA definition of value added is a better representation of
my current understanding of Marx's labor theory of value, though I'm certainly
willing to hear arguments to the contrary."

I don't accept this. First, again, I would NOT use the above to define "the"
monetary expression of value. If we take "p" to be the unit money price, as
Duncan is now doing, then I would "define" the MEV as follows:

(1/m[t])p(t)X(t) - (1/m[t-1])p(t-1)aX(t) = l(t)X(t)


m[t] = p(t)X(t)/{(1/m[t-1])p(t-1)aX(t) + l(t)X(t)}

where the numerator on the RHS is the money value of output and the
denominator {.} is what I understand to be the labor-time value of output.
(This holds even if the above are considered as vectors, i.e. multiple
sectors.) Thus, I understand the MEV at any moment to be the ratio of the
money value to the labor-time value of the product. (Actually, however, since
money is spent on and used to value non-produced assets, I think that, in a
*real* economy, the numerator should be the money value of *all* alienable
assets and the denominator their labor-time value.)

Hence, the real issue is whether (1/m[t-1])p(t-1)aX(t) + l(t)X(t) is indeed
the labor-time value of output, i.e., whether the first term is Marx's C and
the second is his V+S, in labor-time terms. Simultaneist interpretations deny
that the first term is C; the TSS interpretation affirms this.

Second, I deny that my actual procedure, as I have just outlined it, imputes
price changes in stocks to the value added by living labor. Again, I agree
100% with Duncan that such price changes (in labor-time terms and/or in money
terms, it doesn't matter) have nothing to do with production and should
therefore not be imputed to the expenditure of living labor. Look again at
my gross value of the product in labor-time terms:

GVP(t) = (1/m[t])p(t)X(t) = (1/m[t-1])p(t-1)aX(t) + l(t)X(t)

The first RHS term takes the money value of the inputs when they entered
production and deflates them by the MEV of that moment. One may, if one
wishes, say that this "revalues" this stocks, since they are valued (in both
labor-time and money terms) at their price when they entered production, not
the changed value they have after production. I would prefer to say that one
is revaluing them if one uses their changed prices instead of their original

But this is just terminology. In any case, what I claim is that {(1/m[t])p(t)
- (1/m[t-1])p(t-1)}aX(t) is not attributable to the value added by extraction
of living labor in capitalist production, and therefore should not be counted
as part of value added. Therefore I do not do so, but count value added as
l(t)X(t). For instance, if the value of seed-corn when planted is
(1/m[t-1])p(t-1)aX(t) = 500 labor-hours, and workers work l(t)X(t) = 100
labor-hours, I maintain that in Marx's theory, GVP in labor-time terms is
(1/m[t])p(t)X(t) = 600 labor-hours. If, at the end of the production period,
the same amount of identical seed-corn has a value of (1/m[t])p(t)aX(t) = 480
labor-hours, I do not attribute this change in the value of seed-corn during
the year to living labor. I do not say that value added is 120 labor-hours.
I say value added is 100 labor-hours. What I still am very unsure about,
because the use of money sums, accounting conventions, and the word "revalue"
are confusing a simple issue, is whether Duncan agrees that {(1/m[t])p(t) -
(1/m[t-1])p(t-1)}aX(t) = -20 labor-hours is (a) not attributable to the value
added by extraction of living labor in capitalist production, and (b) should
not be counted as part of value added IN LABOR-TIME TERMS.

Note that NONE of this has anything to do with money value. It all concerns
the determination of value, including value added, as measured in labor-time.

Incidentally, my GVP equation differs from what Duncan thinks my equation is.
Using *my* concept of "money value added" -- an expression I don't like, but
will accept for the moment --- and *Duncan's* formula relating the MEV and
money value added (that he thought we both accept), one would get the
following gross value of the product in labor-time terms:

GVP(t) = (1/m[t])p(t)X(t) = (1/m[t])p(t-1)aX(t) + l(t)X(t)

which differs from mine in that the first RHS term is being deflated by m[t]
instead of m[t-1]. I think this is clearly wrong, since m[t] corresponds to
p(t), but m[t-1] corresponds to p(t-1).

Duncan: "When I solve the equation using the NIA definition of Money Value

p(t)X(t) - p(t)aX(t) = ml(t)X(t)

with the same rate of increase in labor productivity, I (I think correctly)
get a price path that declines at the same rate as social labor productivity,
and leads to a constant money rate of profit."

This implies that Duncan's GVP in labor-time terms is:

GVP(t) = (1/m[t])p(t)X(t) = (1/m[t])p(t)aX(t) + l(t)X(t).

Value added is being attributed solely to living labor, as in my equation, but
the first RHS term counts C differently, not due to any monetary factor, but
in labor-time terms. *This* is where we differ. Duncan's formula uses the
post-production labor-time value of means of production, instead of the sum of
labor-time laid out as constant capital at the start of the production period.
But I'm still not sure that Duncan *wants* to deny that the constant capital
laid out at the start of the period is transferred to the value of the
product, as all simultaneist interpretations do.

Thus --- here is my brief comment on ope-l 3288 --- I use the term
"simultaneist" not only to refer to the proposition that the money price of
the inputs must equal the money price of the outputs, but also that the
labor-time value of the inputs must equal the labor-time value of the outputs.
A lot of people are happy to let the money prices vary. But the TSS
interpretation denies that, in *labor-time* terms, the value of constant
capital is determined simultaneously with the value of the output.

Duncan: "1) Do different consistent predictions as to the path of prices and
money profit rate arise in the same interpretation of the LTV depending on the
concept of money value added adopted?"

I would say that the only concepts of "money value added" consistent with the
TSS interpretation are ones that are consistent with the equations I have
presented here.

Duncan: "2) What are the strengths and weaknesses of these two concepts of
money value added in representing Marx's thinking on the relation between the
LTV, technical change and the path of the money profit rate in capitalist

"3) What are the strengths and weaknesses of these two concepts of money
value added in understanding the actual paths of prices and money profit
rates in real capitalist economies?"

Nothing in the concepts or equations I have put forth implies, by itself,
anything about the determination of the path of MONEY prices or MONEY profit
rates, just as the equations of Walrasianism and Sraffianism do not. Nor do I
think Marx made any claims in this regard. He almost always obtained
conclusions by holding the value of money or MEV constant.

That is not to say that these issues are unimportant. They are important.
But the law of value alone will not solve them. In the end, we must be able
to say whether or not, and if so, how, labor-time magnitudes have an influence
on the operation of the system. But this doesn't necessarily mean that
falling labor-time values have to lead to falling money values. If money
values remain constant or rise when labor-time values fall, as has been taking
place for a half century, it might be important to consider the *displacement*
of crisis tendencies. For instance, is debt crisis related to the attempt to
keep up money prices when labor-time values are falling? At this point, I
think, no one can say for sure whether or not labor-time value magnitudes have
an influence on the operation of the capitalist system. Marx could be wrong;
then again, he could be right. Ignorance is not sufficient reason to reject
his theory, or accept it.

None of this, of course, has any bearing on which interpretation of Marx's
value theory, if any, is adequate to the original.

Moreover, the issue of the tendency of the profit rate, and the "definition"
of all value magnitudes, in *labor-time* terms is still far from being
settled. Simultaneism and temporalism differ on the correct interpretation of
Marx's value magnitudes, including value added, in *labor-time* terms.
Simultaneism cannot replicate Marx's conclusions concerning the relations
determining the profit rate (or anything else) in *labor-time* terms;
temporalism can. And I'm still very far from being clear about what Duncan
things about these matters. It would be a shame if we let the discussion of
money obscure or deflect us away from the problem of working out an adequate
interpretation of the labor-time value magnitudes (or constant-MEV value
magnitudes) in Marx's work.

Andrew Kliman