[OPE-L:1012] Re: Money and labour

Gilbert Skillman (gskillman@mail.wesleyan.edu)
Thu, 8 Feb 1996 14:33:08 -0800

[ show plain text ]

Thanks to Alan for his post on this topic. I hope to use it to show
the translational difficulties I'm having with his arguments. In my
understanding, Alan's argument depends on reifying terms which have
solely derivative status, with the consequence that the tautology
I've spoken of in earlier post is brought in through the back door.
In what follows I'll try to spell this out. Alan writes:

> I have two or three preliminary questions to deal with meanwhile
> on which I will post separately.
> The first concerns the supposed 'dimensional incompatibility'
> between price and value.
> In OPE 960 dated 6/2/96 Gil says
> "Putting aside for the moment that L() insofar as the entity
> is commonly understood, is measured in units of labour time,
> and p(), insofar as the entity is commony understood, is
> measured in money unts..."
> I'd rather we didn't put this aside for very long, because it
> is the source of much confusion. The idea of a 'dimensional
> incompatibility' between price and value has been floating
> around for a while...

I didn't say that there was a "dimensional incompatibility" between
price and value, as one can easily verify from the passage above Alan
extracts from my post. I said that the TSS expression of value mixes
together terms which are commonly measured in money units with terms
that are commonly measured in labor units. Thus, one should expect a
conversion factor on either p or L, but there isn't one. This leads
to confusion, as I show below.

Now, of course, there is no "dimensional incompatibility" between
prices and the monetary expression of values, since both are in money
units. However, as we'll see in a second, there is necessarily such
an incompatibility between prices and values as Marx specifies the
latter; thus any direct comparison requires that the latter must be
translated into their monetary expression.

> Marx, in Chapter I of Volume I, explains that both price and
> value have two measures:
> (a) the intrinsic measure, labour time
> (b) the extrinsic measure, money

> There are many citations relating to this which I can provide
> if this idea is disputed but I won't give them here to avoid
> clogging up this post.

I think rather that Marx understands each entity to have two
*expressions*, as opposed to measures. I could also provide
citations where Marx distinguishes between value and its money
expression, but I'll withhold them for similar reasons.

What is quite clear is that Marx understands value to be measured
solely in terms of labour time. He gives two tests (I, p. 130):

"Commodities which contain equal quantities of labour, or which can
be produced in the same time, have therefore the same value."

"The value of a commodity would therefore remain constant, if the
labour-time required for its production also remained constant."

I note that the TSS expression for value, v = pa + l (suppressing
time subscripts), violates both conditions. Two commodities which
can be produced in the same time may not have the same value due to
variations in pa (because, say, they use different input
commodities); alternatively, a commodity's value may change due to
changes in pa, even if the labour-time required for its production
remained constant; this is true even if one controls for inflationary
effects by normalizing the price of the money commodity to one.

I note further that Marx never retracts either of these tests of the
measure of a commodity's value, so that the TSS expression is
necessarily inconsistent with Marx's account. I don't regard this as
much of a problem, however, because as I've mentioned earlier I
understand Marx to put forward two mutually inconsistent hypotheses
about the connection between prices and values. The TSS system,
suitably interpreted, is one method of repairing the inconsistency.

> Thus in the disputed Chapter V discussion of wine whose value
> is $40 and corn whose value is $50, we may represent this equally
> in either money or labour.

I think it's less potentially misleading to say that in this case $40
and $50 represent monetary expressions of the respective commodities'
values. I realize Marx did not use this qualifier in every instance,
and I've offered an explanation: Marx understands price-value
equivalence to be the "pure" case of commodity exchange, on grounds
we now know to be problematic. On the basis of this understanding,
Marx often uses values and monetary expression of values

> If the value of money is 1 hour per dollar or, which is the
> same thing, if the monetary expression of labour is 1 dollar per
> hour, then the wine has a value of 40 hours and the corn a value of
> 50 hours. If on the other hand the value of money is 2 hours
> per dollar then the value of the wine is 80 hours, and so on.
> But exactly the same applies to price. If the price of the wine
> is $45, and the value of money is 1 hour per dollar, then the
> price of the wine is also 45 hours.

There are two problems with this passage, and they are the source of
a good deal of confusion. First, Alan does not define "the value of
money", and consequently does not define its inverse "the monetary
expression of labour." One definition of "the value of money", which
I'm going to use hereafter for the purpose of illustration, is the
ratio given by aggregate prices P = p*x relative to aggregate direct
labor L = l*x. The latter is equivalent to total values as defined
by v = va + l; see Chapter 1 of Morishima, Marx's Economics.

In preparation for the second point, let's define some terms:

Let v(m) = the value of money = P/L, measured in dollars per hour;
m = the monetary expression of labor = 1/v(m) = L/P, measured in
hours per dollar; p(i) = price of goods i, measured in dollars (following Alan I
sub-aggregate across individual goods (i); v(i) = value of goods i,
measured in labor hours, following the passages from Marx cited
above; m(i) = monetary expression of values = mv(i); and z(i) =
expression of price in labor hours = v(m)p(i).

Note is that in general v(i) does not equal z(i)
and p(i) does not equal m(i). To assert otherwise is to impose
price-value equivalence, which we've insisted Marx does not assume,
at least through Ch. 5. Thus, where Alan speaks of 2 terms, there
should actually be 4. In reducing 4 to 2 **he is assuming what must
be proved**. For example, changes in demand conditions in the sphere
of simple circulation may change p(i) and change the labor expression
of price, **even controlling for nominal price changes by setting the
price of the money commodity always equal to one.** I've argued this
in previous posts, but apparently without effect, since Alan writes:

> This is precisely the reason that, even though because of a
> monetary inflation, or what Marx terms a 'general rise in all
> nominal prices', the price of the wine may be $45, or $45*2, or
> $45*3 as Gil says, nevertheless its price when represented in
> terms of the intrinsic measure is always 45 hours, which is how
> the equality of total price and total value holds: the wine's
> value in hours is 40, the corn's value in hours is 50, and the
> price of each in hours is 45 and this is true independent of
> the money price; it abstracts from monetary inflation.

This statement is doubly wrong. First, **even if one controls for
monetary inflation** by setting the price of the money commodity
equal to one, prices can be made to vary arbitrarily by changing
demand conditions. Since p(i) can vary arbitrarily, so will the
labor expression of price, z(i) = v(m)p(i), so it is not true that
"its price when represented in terms of the intrinsic measure is
always 45 hours". Now, v(i) will not vary, since it is constructed
independently, but in general v(i) is not equal to z(i).

Thus, Alan's conclusion "nevertheless its price when represented in
terms of the intrinsic measure is always 45 hours, which is how the
equality of total price and total value holds" is invalid. Rather,
total prices and total monetary expression of value are guaranteed to
be equal IF a) v(i) = v(i)*a(i) + l(i) and b) v(m) = L/P. But this
aggregate equality is not guaranteed by the TSS value expression
v(i)(t+1) = p(t)a(i)(t) + l(t), since prices can change arbitrarily
across periods *even if one controls for inflationary effects by
setting the price of the money commodity = one*.

> We can always express any money price as the product of two
> completely independent factors; the intrinsic measure of the
> value 'commanded in exchange' by a commodity (a magnitude
> expressed in labour hours), multiplied by the monetary
> expression of labour itself (a ratio, measured in dollars per
> hour) to obtain the extrinsic measure of price (a magnitude
> expressed in dollars). It follows that we can break down any
> *change* in money price into the sum of two components: the
> difference between the two intrinsic measures of price, plus a
> factor due to the change in the value of money.

I have two problems with this statement. First, the value "commanded
in exchange" by a commodity, if it's anything, is Vp(i)/P, i.e. the
portion of total independently defined value given by the ratio of
individual commodity price to aggregate price, and there is nothing
"intrinsic" about this measure, since it necessarily depends on
market price. Second, since we don't know how the value of money is
defined, we don't know how the monetary expression of labor is
defined, and so we don't know how it changes. Thus it is at best not
clear what this statement says.

> It follows that, since value in money terms is no other than a
> special price,

Call it that if you want, but it isn't in fact a price, since
commodities do not in fact exchange according to m(i) but according
to p(i).

>we can always express a price-value difference
> as the sum of an intrinsically-measured difference and a factor
> due to the change in the value of money.

This creates an additional level of confusion "price-value
differences" would exist even without any change in prices; thus it's
not clear how *this* conception of price-value difference relates to,
say, v(i) - z(i) or p(i) - m(i) defined for a given price regime.
> I will return to this important point later because it relates
> directly to Marx's definition of the value of money which, I
> agree is not unproblematic, but on the other hand is clearly
> and extensively theorised by Marx and is, in a certain sense,
> his most distinctive economic contribution; his theory of money
> remains not only totally distinct from any other I know of, but
> empirically far more accurate and - I think - the real key to
> his understanding of what capitalism really is. I think I will
> be able to show that this is both consistent and clear.

I agree that this is important for the arguments advanced above; I'll
look forward to this discussion.

> For now, note that every value can be measured either in money
> or labour-time, and *also* every price can be measured either
> in money or labour-time.

We've seen above that this way of putting it creates fundamental
confusions. I think it's safer to say that value can be stated in
its (original) hour measurement or expressed in money terms
(recognizing that this is different than price in general) and that
price can be stated in its original (money) measure or converted into
an expression in labor time which is not generally equal to its
value. The key point is that there must be a conversion from values
as defined into dollar terms, or prices as defined into labor terms,
and this factor of conversion must be made explicit. I note that the
TSS expression for value does not do this; thus the basis for my
comment Alan cites at the beginning of his post.

> These are two different scales for measuring the same thing,
> arising from the fact that the commodity participates in two
> distinct equivalence relations. Relation 1, labour time,
> concerns the circumstances of its production. Relation 2, money
> price, concerns the circumstances of its exchange.
> This is, it seems to me, not an invention of Marx but very
> common in science. For example if I ask you 'how much beer is
> there in this glass' you can answer in two ways: you may either
> say 'a pint' or 'twenty ounces'. Neither answer is wrong; and
> the relation between the two measures is an important
> scientific ratio, density.

This is not a good analogy. 20 ounces = 1.25 pints always and
forever. However, it is not the case that p(i) = mv(i) for all i,
unless one assumes price-value equivalence, which we know Marx does
not do in Volume I, at least through Ch. 5
> Of course if we ran a pub in which the customers asked for four
> whiskies, and we gave them two glasses plus two grammes of
> whisky and said 'that makes four', we would get into trouble.
> If we gave them two glasses plus two Kilogrammes, they would
> get into trouble. Either way, we get trouble.

And exactly the same problem arises in the expression v = pa + l,
recalling that in general it is not the case that p(i) = m(i) or v(i)
= z(i). Thus there must be an explicit conversion factor for either
p or l in order to avoid confusion.


> In the equation you give in OPE-L:960, namely
> V = [p(t)A(t) + L(t+1)]* x(t+1)
> all quantities must be in the same unit, viz either dollars, or
> hours.

But they aren't, because for that to be the case either p(t) must
explicitly be converted into its expression in hours, z(t), or L must
explicitly converted into its money expression, Lm. This is not

> However, if we use dollars instead of hours, there is an
> additional problem to deal with if the value of money has
> changed between [t] and [t+1]; we must make a correction for
> inflation. That is, we must distinguish between current prices
> and historic prices, just as the National Income Accounts do.
> Otherwise, we will find that enormous profits can be made
> without producing anything, as follows: suppose the value of
> money halves between 1990 and 1991. Then, if I buy a bar of
> iron in 1990 for $100, and merely hoard it, then in 1991 the
> same bar of iron will be worth $200 and I have a money profit
> of $100 by doing nothing.

This is not at all an issue in any post of mine, because in every
case I've abstracted from inflationary effects by setting the value
of the money commodity equal to 1.

> I know I've asked this before but I can't resist doing it
> again: under these circumstances, do you consider that new
> value has been created?
> And do you really think it is wholly unreasonable, circular
> or without foundation, to claim that no new value has been
> created?

And I can't resist pointing out that this has never been the issue.
Of course new value, defined as socially necessary labor time
expended in production, is not created in exchange. Rather the
question is whether total prices necessarily equal the (monetary
expression of ) total values. They don't in general, unless one
defines the value of money so as to guarantee this.

> A much more serious problem, therefore, than the supposed
> incompatibility between hours and dollars, is the very real
> incompatibility between 1990 dollars and 1991 dollars which you
> seem to ignore.

What? Not only have I never ignored it, I've always controlled for
it in my argument.


> But the simplest and most general way to resolve this is to
> reduce all magnitudes to a common measure by expressing them in
> the intrinsic measure of value, labour hours. This is the
> simplest way to interpret the equation you supply. Therefore,
> everything in this equation should be given in hours.

But then the price term p(t), which is given to us in money terms,
must be explicitly converted to z(t) by using the conversion factor

In solidarity, Gil