[OPE-L:7103] [OPE-L:601] Jam sandwich: a response to Steve's [OPE-523]

Alan Freeman (a.freeman@greenwich.ac.uk)
Fri, 05 Mar 1999 14:36:07 +0000

In the flurry of posts I've missed a couple of responses I owe Steve, in
particular the first part of [OPE-L 523] which I think is important. If I've
missed others that he considers important perhaps he can remind me.

Steve wrote

>Your axiom of basket decomposition rules out complementarities between
>exchanged commodities. (snip) Following your axiom (snip),
>if 2 books exchange for 3 bread and 1 socks exchange for 3 jam, then 2
>books and 1 socks would exchange for 3 bread and 3 jam. This is unlikely
>as bread and jam are complementary and therefore together would demand more
>books and socks in exchange than they would in the antecedent exchanges.

Alan replies:

It was when I read this that I started to understand more clearly what's been
going on in this discussion. Andrew's clarification concerning the
commodity-relation also helped.

Bread and jam are complements when they occur in a jam sandwich. This is a
distinct use-value, not an aggregate of existing use-values. Oxygen and
Hydrogen are 'complements' in that they make water, but the exchange-value of
water cannot be deduced from the exchange-value of these gases.

A jam sandwich has a higher exchange-value than separate jam and bread because
it embodies the labour of sandwich-making, not because jam and bread are
complements. Water is cheaper than bottled Oxygen and Hydrogen because
labour is needed to break it up, not to put it together.

You can't use the concept of complement to deduce exchange-value; it doesn't
tell you anything and it leads to contradiction, such as the one I just

I think what you are trying to do is retrofit essentially neoclassical
concepts of exchange to Marx's concept of the commodity. You seem to wish to
specify exchange in terms of what people are *willing to pay*, not what
*actual exchange relations* must prevail. I disagree for two reasons, of which
so far I've only given one

Reason (a) : Marx didn't use neoclassical concepts to explain exchange-value;

Reason (b) (New): neoclassical concepts don't explain exchange-value anyhow.

They are *inferior* concepts.

Thus I think what is being done is doubly wrong. In particular, you can't use
an *inferior* concept to judge a *superior* one. At the very least, you have
to demonstrate it is superior before you begin using it in this way.

In fact Marx's commodity-concept can be used to explain many of the phenomena
you cite as refutations of this commodity-concept, and I'm happy to
demonstrate that. But the reverse is not the case. Neoclassical concepts
cannot explain actual exchange, not even the simple facts of commerce.

Let's illustrate this. If laws of commodity exchange could really be governed
by the principles you suggest, then supermarkets would charge as follows:

(a) does the customer have either a jamjar or a breadloaf in the basket?
(b) if so, check for the complement.
(c) if the customer has either jam or bread, but not the other, charge the
barcoded price
(d) otherwise, add a surcharge.

No supermarket I know could get away with this. A discount maybe, but a
surcharge? Not.

Moreover, this failure of your suggested concept doesn't arise because of
some exceptional or odd circumstance, some freak of the market. It arises
when markets function exactly as they are supposed to function. It arises
out of normal commerce. Normal commerce is incompatible with what you
propose as the definition of exchange.

That is, your definition of exchange is simply unsuited to describe normal

What does this establish? That the relation of 'exchange' which you are
mentally assessing could not possibly govern the exchange of *commodities*.

Now, does this mean it's not a valid or useful concept of exchange? No,
probably it is a valid concept of exchange. So is, for example, an exchange of
favours, an exchange of vows, and exchange of views, etc etc

But exchanges of this nature do not take the commodity form. To be precise,
the commodity relation imposes additional laws, laws of exchange that are
specific to the commodity form. I think it would be useful to seek to specify
what these are. Among them I think we would find the following:

(a) Non-subjectivity or 'sociality' to coin a phrase. If aRb for one person
then aRb for another. aRb is a social relation, not a purely private relation
between individuals. Indeed, this is why we can simply state 'aRb' and not
have to qualify it by saying 'my brother's preferences reveal that for him,

(b) The indecomposability of use-value. A jam sandwich is a distinct
use-value; it isn't the simple aggregation of the use-value of a jar of jam
and the use-value of two slices of bread. Apple and flour are distinct from
apple pie; coffee and milk are distinct from caffe latte, and so on. The
'terms' in the commodity-relation are use-values, not atoms.

I wouldn't dignify these propositions as axioms; they are more
metamathematical, relating to the 'prior analytics' of the discussion: they
clarify the kind of objects we might be talking about. Both seem to me to be
inherent in the nature of the commodity; most of the counter-examples I've
seen refer to types of exchange, however, where one or other of them is
bypassed or violated.

I'd like to reiterate my earlier view. I think that concepts of
exchange-equivalence based on indifference, preference, and so on, can only
logically apply under equilibrium. They cannot be the basis of actual
exchanges for the precise reason that Gil admits; namely, arbitrage would blow
them away.

Of course, to the extent that the full commodity-relation does not apply,
arbitrage takes place, as an *exception* to the commodity relation, as a
modification of it, a relaxation of it. In this respect (only!), Marx's
concept of commodity-relation has a status no different from neoclassical CGE;
it's an initial abstraction.

One must then ask, however, which is the better abstraction. Not all
abstractions are equally valid.

One must in particular ask: can the abstraction be relaxed to deal with
concrete violations (two prices in distinct national markets, price
discrimination, etc)?

My brief answer is the following: Marx's abstraction, the commodity-relation,
can be so relaxed, can accomodate
violations and deviations, because its assumptions are sufficiently minimal
(and in particular because it does not assume equilibrium) to accomodate
infinte variety. The neoclassical abstraction, general equilibrium, cannot be
so relaxed, cannot accomodate violations and deviations, because it
presupposes an entire configuration of all its elements which hardly, if ever,
exists, and which rules out change.

That's why I consider the commodity-relation a better basis on which to study
actual markets than neoclassical general equilibrium: it's a better