I will admit to Gil that the word "marginal" has a similar effect on
me to that which the word "culture" had on Goerbels (sp?). The reason
for this reaction is, of course, a complete rejection of the neoclassical
paradigm where such notions are paramount, and where the "twin" to
marginal cost is marginal (subjective) utility. I see Marx as a
continuation of the "just-price" position, where the costs of
production were regarded as the basis for the setting of price, with
fluctuations in "consumer preferences" leading to quantity changes,
but not determining the price.
If Gil has a similar perspective on Marx, then maybe we should stop
arguing now, because our differences may well be "marginal" -:).
However, I still see Marx's analysis has being more consonant with
economic systems in which there is excess capacity in both labor
and capital, with the former existing for reasons that Marx
elucidated well, and latter existing for reasons that are covered
by Post Keynesians such as Kornai, in his concept of "demand-
constrained economies". In such economies, the firm's costs are
generally constant over the relevant range of output, and if the
firm finds demand pushing it into a range where it suffers from
diminishing marginal returns, its response is normally to (a)
outsource to firms with excess capacity, (b) construct additional
capacity as quickly as possible.
In such industries, average cost determines the lowest cost
producer, whether that is due to low fixed costs or low marginal
costs, or indeed to non-reproducible quality differences.
I must also admit to warning bells flashing when I see such concepts
as "long run marginal cost curves" bandied about. In neoclassical
economics, such curves are drawn assuming constant technology...
need I say more? If you mean a more realistic concept with
technology changing through time, then you are going to have as
the rule falling "long run average cost curves" in real terms for
most firms. If technological development is seen as a fixed cost,
or partly fixed and partly variable, but nonetheless the determinant
of how much one firm's LR cost curve slopes down compared to another,
then the issue of "marginal cost" becomes irrelevant in the long
run too. Particularly so if we also allow for "path determined
futures", where the firm that invests more in R&D grows faster...
The concept of marginality is also normally applied in a timeless
guise. I realise that I may be accusing Gil of something for which
he is not "guilty", and if so I apologise. But the entire notion
of marginal costs is so tainted with the neoclassical notions of
timeless optimisation that I would want solid justification for
its superiority _in a time-based setting_.
Cheers,
Steve