This started out as a reply to Duncan's ope-l 3128. The quote in the next
paragraph is from that, but thinking about things as I wrote led me away from
the BEA and IVA, and the post got very long without even returning to that.
So this is really a bunch of additional thoughts about the value added issue
in general.
Duncan: "revaluation of inventories due to price changes during the
accounting (or production) period is not part of the value added to the inputs
by production itself (or as Marx would say, by the expenditure of living
labor), so it shouldn't be counted in the value added."
I agree with this 100%. As I use the term "revaluation," simultaneous
determination of input and output values revalues the used up means of
production, temporal determination does not. To return to my example in ope-l
3097:
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
etc.
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
etc.
The value of the seed-corn at the end of year 0 = start of year 1 is 500
labor-hours. To this I add 100 newly expended labor-hours, to arrive at 600
labor-hours as the value of output at the end of year 1. Simultaneous
determination revalues the seed-corn down to 400, though not in a
straightforward manner that Duncan employed in responding to John. Rather,
here is what is done: Year 1's output is 6.25 bu.; seed corn input is 5 bu.
The unit input value = 500/5 = 100; the unit output value = 600/6.25 = 96.
Revalue the total input value down to 96*5 = 480, so total output value is now
580. But this makes the unit output value = 580/6.25 = 92.8. Revalue the
total input value down to 92.8*5 = 464. Etc.
So the simultaneist "C" is what the inputs *would have* cost *if* they had
been purchased at the value that the output *would have* had *if* the output
had been produced by means of inputs that had the value that the output *would
have* had. Or, by the Morishima-Skillman-Cockshott-Cottrell interpretation,
"C" is the labor-time that *would have* been needed to reproduce inputs *if*
the labor-time needed to reproduce them were what the labor-time needed to
reproduce the outputs *would have* been *if* the outputs had been produced by
means of inputs that required the amount of labor-time that the outputs *would
have* had.
Ouch. My head is starting to hurt. This procedure removes valuation from the
realm of real processes of production (and circulation) and displaces it to a
virtual reality in which values never refer to actual costs incurred. (Note
that for values to refer to actual costs *incurred*, output values and prices
don't have to be equal.) This does not bother some people, but it bothers me,
and I'm almost certain that it bothers Duncan. I infer from Duncan's
publications and posts that he finds his interpretations and/or concepts of
the value of money and the value of labor-power fruitful largely because they
give value theory purchase over real processes and real data. I fully agree.
Hence, that simultaneous determination implies that commodities have, at one
and the same "time," different values as outputs and as inputs is symptomatic
of an even larger problem, IMO.
Now, in Duncan's recent posts, especially his replies to John, he does *not*
use the method of simultaneous determination. Rather, he straightforwardly
takes the *actual* output price and subtracts what the inputs would have cost,
had they been purchased at the actual output price, in order to obtain value
added. Applied to the above example, *if* the TSS value of output in year 1,
600 = 96*6.25, is the actual value of output, Duncan would subtract 96*5 = 480
and obtain a value added of 120. I fail to understand why this is not a
revaluation of the seed-corn. It was worth a total of 500 at the start of the
year, and its value is being reduced by 20 to arrive at 480. This -20 is not
part of the value added (subtracted?) by the expenditure of living labor, so I
don't think this -20 should be subtracted from value added.
But maybe I don't understand what Duncan meant by the statement I quoted
above.
Duncan's value added formula has some other problems. First, I think he
*wants* to use *actual* prices (values) to determine value added. Value added
per unit is, by Duncan's definition, VA = l = p(t) - p(t-1)a - [p(t) -
p(t-1)]a = p(t) - p(t)a, where p(t-1) and p(t) are *actual* prices (values), I
think. But this implies that p(t) = p(t)a + l = l/(1-a). (a and l can change
over time.) The right-hand-side indicates that Duncan's formula is not merely
a method of *measuring* value added, but implies very specific relations of
*determination* of output values. Unless the *actual* p(t) is always equal to
l/(1-a), the formula is self-contradictory. So we're back to simultaneous
*determination* of input and output values and its unbridgeable split between
value theory and reality. This is not what I think Duncan wants.
Second, depending on how one interprets it, Duncan's argument that the TSS
interpretation miscalculates value added may rest on a petitio principii, beg
the question. He wants to show, I think, that TSS value added doesn't equal
living labor. But to do so he employs his value added formula, which
*presupposes* that TSS input and output values, and therefore value added, are
all wrong.
(There's also another difficulty with using Duncan's value added formula to
test whether TSS value added equals living labor that I will develop below.)
On the other hand, Duncan's demonstration in his reply to John may have been
intended to show that the TSS concept of value added differs from the BEA's.
I'm more confused than ever about this, but willing to accept it.
Third, unless Duncan is claiming that the simultaneously determined prices
(values) are the *actual* ones, as I noted a few days ago, he seems to lack a
way of determining the output prices (values). In itself, I think this is
unimportant. And, for some problems, one can use actual data. However, two
associated problems are important, IMO. One is that a theory of the
determination of profitability and a theory of the determination of the
aggregate price (value) of output go hand in hand. Simultaneism deduces the
profit rate from the condition that input prices equal output prices. Marx
takes the results of production to determine the profit rate and therefore
aggregate prices.
The other is that, if we want to relate value categories to actual data, we
need to correct for changes in the monetary expression of value (MEV). But to
get the latter, we need to know the constant-MEV values. Depending on how
they are determined, the results will differ. Take my year 1 labor-time
numbers. Assume that 1 labor-hour is expressed as $1 at the start of the
year, but $1.10 at the end. Hence, in money terms, the simultaneist C+V+S =
$550. Since the corn input = 0.8 times the corn output, Duncan's value added
calculation gives a monetary value added of $550 - 0.8*$550 = $110, so value
added per labor-hour = $110/100 = $1.10. This corresponds to the end of the
year MEV, which is, I think, what Duncan wants.
But unless the simultaneist "price is right," the figures don't add up.
Assume the *actual* value of output is $551.10 in money terms and 501 in
labor-time terms (again, 1 labor-hour = $1.10). By Duncan's value added
procedure, monetary value added = $551.10 - 0.8*$551.10 = $110.22, so that
monetary value added per labor-hour = $110.22/100 = $1.1022. This is larger
than both the initial $1 and the final $1.10, so something seems wrong.
Using it, we deflate the money value of output to 551.10/1.1022 = 500
labor-hours, so 1 hour is unaccounted for. (The revalued C will equal 400
labor-hours and thus living labor will equal 100 labor-hours, so there's no
problem with value added in the narrow sense.)
Thus, the correct procedure for obtaining value added depends on how value is
*actually* determined. Duncan's procedure works for values that are actually
determined as l/(1-a), but not otherwise. In other words, there is a one to
one correspondence between this procedure and that theory. So, unless I'm
missing something or making an error in calculation, I think Duncan needs
either to accept simultaneous *determination* of input and output values, and
all it implies, or to accept an alternative procedure for assessing value
added in the general case.
Again, I agree that if the TSS interpretation produces a divergence between
living labor and value added, it is wrong. But the above indicates that
Duncan's value added equation really shouldn't be used to test this. I would
be persuaded if one could deduce a divergence beginning from the *concept* of
value added itself or from the TSS equation itself or it one could show the
TSS equation to be incompatible with the texts.
Andrew Kliman