[OPE-L:3097] RE: Value added, IVA and TSS

andrew kliman (Andrew_Kliman@msn.com)
Mon, 23 Sep 1996 12:33:08 -0700 (PDT)

[ show plain text ]

I think Duncan's ope-l 3074 closed the book on an earlier phase of the
discussion on Marx's law of the FRP and at the same time opened up a new one.
It is because it opened up so many issues that this response is necessarily
incomplete and preliminary.

I very much appreciated Duncan's post, partly because of his forthrightness
and honesty, and largely because it represents a significant advance in the
process of clarifying what the differences are. I think we are in agreement
at this point that the differences between TSS and simultaneist value theory
are not reducible to particular assumptions of particular models, but concern
the concept of value. I remember Alan's excitement when Riccardo recognized
what we were saying about the Ch. 9 transformation, and declared "Somebody
Finally Got It!" I feel the same way about Duncan's recognition that we
differ on elemental matters of value theory.
This lets us put aside, at least for the moment, the evaluation of the
assumptions of particular "models" (though I see John is minding his manners
by writing that wages are "next to nothing" and "v is very,very,very small")
and lets us focus on the conceptual issues.

BTW, since Duncan mentioned that he now understands better why people have
trouble accepting the TSS position, I want to reiterate that I am not
interested in persuading others to agree with it-which is why I can be so
happy that he "gets it" but doesn't agree with it. Instead, I have claimed
an attempted to show that it is an adequate formalization of Marx's value
theory, in contrast to other interpretations, and that it removes the
appearance of inconsistency in Marx's theory. I want others either to show
that these claims are wrong, or to acknowledge that they are right.

The new stage of discussion Duncan has opened up is conducive to this focus,
because now I *think* we are addressing something I *think* we all agree that
Marx claimed: all new value is created by living labor. We are not
discussing any sort of model, but probably the most elemental claim of Marx's
own value theory, and the implications of that claim.

I think Duncan's distinguishing of the "value added" issue from the
measurement of the profit rate is EXTREMELY important and helpful. We are
agreed: historical costs and replacement costs can each be used to measure
different things that, however, can both be called "rates of profit." We
don't have to talk about which one gets equalized, or doesn't; which one
capitalists care about; their expectations; or anything else.

There is no need for verisimilitude. We can ignore fixed capital. We don't
even have to talk about multiple sectors that lead to quantitative price/value
differences. We can (I hope) focus on a one-commodity capitalist economy, and
simply compare the implications of two equations that each claim to formalize
value determination in Marx's theory, given the above conditions:

v(t) = v(t)a(t) + l(t),

which ALL interpretations except TSS (and certain "abstract labor"
interpretations) adhere to in the one-commodity case,


v(t) = v(t-1)a(t) + l(t),

the TSS interpretation.

(v(t) is the unit *output* value = price of period t; v(t-1) the unit *input*
value = price of period t.)

The debate has been deflected away from this elemental difference in the past,
impeding development (I'm partly to blame for this, having taken Duncan
through an example where a = 0, and the difference between these equations
disappears). If we can rigorously stick to it now, I think we'll have a lot
of progress. "We" means the list as a whole. Thanks to Duncan, we have a
clear question-which interpretation is consistent with the proposition that
all new value is created by living labor? The above conditions and resulting
equations give us a simple, clear-cut way of testing it.

I also hope we stick to the equations and their implications, and not return
(yet) to scrutinizing passages from Marx's texts haphazardly.

OK. Now to the nitty-gritty. I find the concept of the IVA (and the concept
of replacement costs, generally) misleading in this context. The IVA and
replacement cost valuation are accounting concepts/techniques meant to deal
with *inflation* (and deflation). For practical reasons, managers want to
know how much of their profit comes from what is, essentially, a nominal
capital gain: inflation raises the money value of the raw materials they use
in production, or goods they resell, over their value when they were produced.
Now, we could discuss whether these capital gains are or are not part of
profit, the history of accounting in relation to this, what Marx said about
it, etc., probably for a long time. I propose we do something more simple:
assume away inflation, and then compare the implications of the
"inflation-adjusted" equations.

This, however, is not as straightforward a proposal as it seems. There are
two different concepts of inflation to consider. The usual one is that if a
given aggregate of use-values costs more money, there is inflation. Marx may
utilize this concept, but he also has a concept of inflation of the monetary
expression of value: a given amount of labor-time is expressed as more money.

Whenever the labor-time needed to reproduce commodities changes, these two
measures of inflation will differ. Imagine, for instance, that at time 0, it
takes 10 units of labor to produce a widget, and that the money price of a
widget is 5 yen. At time 1, it takes 9 units of labor to produce a widget,
and the money price remains 5 yen. Assume that widgets are representative of
the whole economy. Then, by the usual definition, there is no inflation, but
there is inflation of the monetary expression of value. At time 0, 1 unit of
labor was expressed as 5/10 yen. At time 1, 1 unit of labor is expressed as
5/9 yen. The rate of inflation of the monetary expression is 11.11%.

Imagine, conversely, that at time 1, the money price of a widget has fallen to
4.5 yen. Then there is a 10 0eflation, according to the usual concept. But
1 hour of labor is now expressed as 4.5/9 = 5/10 yen, so there is no inflation
or deflation in the monetary expression of value.

Now, I think Marx would say that there has been a real fall in the unit value
of a widget, or a fall in the "real value" of a widget, *irrespective* of
whether the money price remains at 5 yen or falls to 4.5 yen. (I find it hard
to believe that anyone would disagree that this was his view.) The real value
of a commodity falls (rises) when less (more) labor is needed to reproduce it.

This implies that the usual concept of inflation is not appropriate for
discussing the determination of value in Marx's theory. And hence, neither
are the *usual* concepts of "replacement cost" and "IVA," which are closely
related to it. According to the usual concept of inflation, we have a
"nominal" fall in the value of a widget, "deflation," if the unit price falls
from 5 to 4.5 yen. According to Marx's theory, however, there is a "real"
fall in the value of a widget, and, when the monetary expression of value is
constant, this fall is expressed monetarily in a "pure" fashion.

Hence, to distinguish between real changes in values, caused by changes in
labor-time needed to produce commodities, and nominal changes in values, one
must either (a) hold the monetary expression of value constant, or (b) measure
value directly in labor-time. These two procedures give the same answers up
to a constant multiple.

Once this is done, and only then, is it possible to address the very important
issue Duncan addresses:

"it seems to me that the nub of the issue is, ... in terms of the labor theory
of value, whether or not one attributes the change in the money value of
inventories over the production period to the expenditure of living labor."

His own view is that "a(p(t)-p(t-1)) is a purely financial effect induced by
the change in prices."

I hope we can now agree on the following. If, in the above example, the money
value of a widget equals 5 yen both at time 0 and at time 1, then this
NON-change in its money value is purely financial, a "veil," from which one
would be wrong to infer that its real value has remained constant. If,
however, the money value falls from 5 yen to 4.5 yen, then this change in the
money value of a widget is not a "pure financial effect," because the 10 0.000000all
in money price exactly reflects the 10 0.000000all in real value.

If we can't agree about this, then we have a major difference concerning
Marx's texts, which only a discussion of the textual evidence will resolve.

Thus, with respect to the above equations, I am suggesting that we measure l
and v directly in labor-time, or in money figures adjusted so that each "yen"
expresses a constant amount of labor-time. The resulting numbers will give us
the *real value* of the commodity according to two different sets of
interpretations, so that no further adjustments need to be made in order to
eliminate financial effects. (And clearly, variation in *relative* prices is
not at issue in a one-commodity economy.)

Let me go back to my example of the other day. Let's call the commodity
"corn" (how original). In the initial period, year 0, 4 bu. of corn and 100
hrs. of living labor are used to produce 5 bu. corn. In each subsequent
period, output increases by 25%, and the input/output ratio remains 4/5, which
together imply that the *whole* corn yield at the end of one year becomes the
seed corn at the beginning of the next. (Wages and capitalist consumption are
very, very, very, very, very, very, very, very, very, very, very, very, very,
very, very, very, very, very, very, very, very, very, very, very, very, very,
very, very, very, very, very, very, very, very, very, very, very, very, very,
very, very, very, very, very, very, very, very, very, very, very, very, very,
very, very, very, very, very, very, very, very, small.) The total amount of
living labor extracted remains constant, 100 hrs., each year. Assume a static
equilibrium in year 0, so that the unit value of corn as input and as output =
100 labor-hours. The production time is assumed to be one full year, and
the circulation time is assumed to be very, very, very, ... short.

We can also work with money figures IF we keep in mind that the actual money
prices have been adjusted so as to remove changes in the monetary expression
of value. Hence, if I now stipulate that 1 labor-hour = $1, the numbers may
be read as dollar amounts as well as hours, and the dollar amounts do not and
cannot vary because of any financial effect, but SOLELY because of changes in
real value, if value is determined by labor-time.

For simplicity, I won't bother to report the physical quantities, or unit
magnitudes, just *aggregates*.


TSS Interpretation
year C V+S C+V+S
0 400 100 500
1 500 100 600
2 600 100 700
3 700 100 800

We begin with 400 hours of labor embodied in the initial seed-corn to which
100 hours of labor are added. The whole of the output produced at the end of
year 0 becomes the seed-corn advanced at the start of year 1, so the *value*
of the whole of the output produced at the end of year 0 becomes the *value*
of the seed-corn advanced at the start of year 1 (as must be the case, since
the end of year 0 *is* the start of year 1). Added to this is another 100
hours of living labor during year 1, etc. All the value produced remains
within the circuit of capital, none is consumed, and so total value increases
by 100 each year, i.e., by the extra value added by living labor, and only by
living labor, each year.


There are no financial factors involved. We are adding amounts of labor-time,
or dollar figures that reflect a constant relation between money and

The corresponding table if valuation is simultaneous gives us:

All Interpretations Except TSS
year C V+S C+V+S
0 400 100 500
1 400 100 500
2 400 100 500
3 400 100 500

Now, I think this clearly does "violate Alan's principle that the money paid
for the inputs ought to equal the money received for the inputs." The farmers
"sell" their corn for 500 but "buy" it for 400. The moment the ball descends
in Times Square, midnight of Jan. 1 of each year, 100 labor-hours vanishes
into thin air. This is exactly what v(t) = v(t)a(t) + l(t) implies, when a(t)
= 0.8, l(t) = 20*(0.8)^t, and output is X(t) = 5*(1.25)^t, as they are in this

It may not be, however, what Duncan means. I'm not sure. Duncan seems to
argue that the input and output values are not necessarily equal:

"In the equations we clearly distinguish p(t) from p(t-1), so input prices are
not being assumed to be equal to output prices. The equations clearly reflect
the fact that the money paid by the capitalist for the inputs is p(t-1)a, the
same as the money received by the producers. The issue is whether or not in
applying the principle that it is living labor that adds value to the product,
we should count the IVA as part of the value added or not."

But if input and output values are not necessarily equal, then v(t) = v(t)a(t)
+ l(t) is NOT an expression for the determination of the output value.

What, then, is Duncan's expression for the determination of the output value,
if he doesn't say that input and output values must be equal? I do not know.

A related question: what expression for the determination of the output value
"attributes" or "imputes" the whole of the value added, neither more nor less,
to the expenditure of living labor? The living labor expended in each period,
I hope we all agree, is 100 labor-hours. If we say that its (constant)
monetary expression, $100, is the money value added, and if we say that the
initial value to which it is added is "the money paid by the capitalist for
the inputs," which is p(t-1)a times gross output or

p(t-1)*(0.8)*5*(1.25)^t = p(t-1)*4*(1.25)^t

and if we further say that the money value of gross output is p(t) times gross
output or


then we get

p(t)*5*(1.25)^t = p(t-1)*4*(1.25)^t + 100.

And if we start with p(0) = p(1), then we get the TSS table, above. Hence,
given what I think Duncan means by value added, it is temporal valuation, not
simultaneous valuation, that equates value added and the monetary expression
of living labor expended. It is, moreover, not just any temporal value
equation which does so, but only the TSS equation.

Andrew Kliman