[ show plain text ]
Some replies to OPE-L 3888, 3891, 3892, and 3893.
I'm glad Steve Keen and I agree. But, Steve, the "Marxist" variant of my
airplane example did indeed "lead" to the (false) conclusion that
airplanes CANNOT fly. The two variants of the example were NOT static
and "dynamic/Marxian." The first variant used standard comparative
static equilibrium methodology. But a lot of Marxist economists don't
like to think of themselves as static equilibrium theorists, so they tell
the story in terms of "reproducible states" rather than comparative
statics. Thus the second variant, which reaches the same conclusion as
standard comparative statics.
Steve also qualified his agreement with a statement that he does not
"regard the neoclassical response to the Sraffian critique as
legitimate." I don't know what he means by "legitimate," so it is hard
to repsond. I think it is undenaible that neoclassical theory in its
present high-theory form has nullified the reswitching critique and
critiques based on illicit aggregation of capital. But it has paid a
rather high price. It now generates almost no results unless it imposes
bizarre conditions like perfect foresight and complete forward markets
for everything.
Rakesh wrote:
"Now what I think the critics are implicitly thinking is that we are
simply allowing prices to vary any which way over time with no
consideration for the constraints imposed by the reproduction of the
system on how much freedom prices actually have to vary
interperiodically."
Response:
It is true that static equilibrium theorists often present the false
binary of static equilibrium vs. a "disequilibrium" in which anything can
happen and therefore nothing can be said. But when they apply that false
binary to me (I won't speak for others), it is simply a straw man. It
has no basis in anything I've written. In Marx's theory, value cannot be
altered in exchange, so total price in the market is constrained to equal
the total value already produced. There is also the process by which
profit rates *tend* to equality, so price deviations from production
prices are bounded in some weak sense.
The reproduction of the system, however, DOES NOT impose constraints on
prices. That is the fallacy my "Marxist" version of the airplane example
was criticizing. It is just ass-backwards to argue that effects (e.g.,
reproduction) determine causes rather than the opposite. If prices are
*in fact* constrained in some fashion, by some *real* processes that take
place BEFORE reproduction, then (for the sake of argument, and cet. par.)
reproduction will take place as a result. That is all.
In practice, the argument from reproduction is even worse. A prime
example is Roemer's wholly disingenuous sleight of hand. He first says
that if reproduction is to occur, then no stock of any good can be run
down to zero. (That's false -- to my knowledge, there are no stocks left
of whalebone stays for corsets -- but let it pass for now.) Then he
says, correctly, that one way of ensuring that no stock runs down to
zero, i.e., one SUFFICIENT CONDITION, is that there be a non-negative net
output of everything in every period. Then he immediately turns around
and calls this a requirement for reproduction, i.e. a NECESSARY
CONDITION, which it most definitely is not.
A great deal of his book (Foundations) collapses with this "error,"
including his false "proof" that simultaneist "profit" is positive if and
only if simultaneist/dual-system "surplus-labor" is positive (the
"Fundamental Maxian Theorem"). In a paper forthcoming in Capital and
Class, I show that (and how) reproduction can continually take place when
some net outputs are negative, and that, if this is the case,
simultaneist "profit" can be negative when simultaneist "surplus-labor"
is positive and simultaneist "profit" can be positive when simultaneist
"surplus-labor" is negative.
I find it somewhat difficult to respond to Ian Hunt's post, because he
takes me to have been criticizing the "Sraffian (simultaneous equation)
*model*" [my emphasis], while in fact I was criticizing the comparative
static *method* and its "Marxist" variant, the *method* of deduction from
reproducible states.
Futhermore, it is not fully clear to me what Ian means by "model."
He argues that the "Sraffian price system ... is not logically invalid."
Again, I said nothing about this. But let me point out that, when there
exist self-reproducing non-basics with a low own-rate of reproduction
(Sraffa's "beans"), then -- as Sraffa himself noted -- profit rates can
equalize only if input prices and output prices diverge.
Ian's main point seems to be that conclusions about the real world cannot
be validly deduced from models. He writes:
"there is a question of the degree to which it [the Sraffian system]
corresponds to real world commodity pricing.
"It would be logically invalid to conclude from the model that market
prices
in actual market economies work (even with rents included) in the way the
model says."
This was my point as well. The method of comparative statics and the
method of deduction from reproducible states are invalid methods. They
"deduce" conclusions from the mathematics of models in an invalid way.
Valid conclusions can be deduced only with respect to the *special cases*
to which the models refer (e.g., the properties of the static equilibria
themselves), not to the general case. But the methods in question pass
invalidly from the special case to the general case.
I suppose one could attribute this to "human error," and blame the
practitioners of the methods rather than the methods themselves. And I'm
happy to blame them. But the "errors" are, first of all, not isolated
instances but standard practice. Second of all, to blame only the
practitioners and not the methods is rather like saying "guns don't kill
people, people kill people." (This is the traditional line of the
National Rifle Association in the US, which it uses to oppose gun
control.) Yes, people kill people, but they do so with guns, not sealing
wax. And yes, people deduce invalid conclusions, but they do so by
saying in one breath that their models don't apply to the real world and,
in the next breath, applying them to the real world by means of the
invalid methods we're discussing.
Ian: "A Sraffian model does not assume that everything is bought and
sold on the same day."
True, but no one has said the opposite.
Ian: "As I understand it, it assumes that the "most important
constraint" on (medium term) prices is the requirement that they pay for
replacement of what has been used up in production together with a normal
return on those replacement costs."
I have discussed the notion of reproduction as a constraint above. The
same applies to replacement.
"How to cash out "important constraint" is a matter of debate. You could
use an "attractor" model or a boundary model, with the constraint
representing a condition that that separates succesful from unsuccesful
commodity producers over the long run, if conditions remain the same."
I'm not sure I understand this. Taking a stab, let me just say that the
average value of a variable (e.g., average price, production price) is
NOT in general equal to the equilibrium value of the variable (fixed
point attractor, e.g., price at which input and output prices are equal).
If the time path of variable X is X(t+1) = 4X(t)*[1-X(t)]; 0 < X < 1,
then the equilibrium value is 0.75, but the average value is 0.5. The
equilibrium is 50 0reater than the average.
Ian: "As I understand your point (and I may well be missing it), it is
that you
think that this idea is wrong and that we should be looking at capitalist
enterprises as essentially money making, where the return on outlays is
what is crucially important. If that is right, the case should be made
in
that way without too much logical embellishment: after all, it is enough
that the Sraffian model picks out the wrong key feature to model (while
abstracting from all the rest). Saying it is "logically invalid" just
says
you can't infer from a model to reality (which is the case even with good
models, or evenwith any of Allin's, which will not actually represent
relative prices at any time in a commodity producing economy)"
Well, I have taken the liberty of making more than one point. Sometimes
my point has been that conclusions about the real world are being deduced
invalidly from special cases, using improper methods. One example of
that, again, is Roemer's alleged "Fundamental Marxian Theorem." Another
is the Okishio Theorem, which invalidly concludes, on the basis of a
comparative-static deduction from a special-case (input price = output
price) model, that Marx's law of the tendential fall of the *real world*
profit rate is false.
Since my point in those cases is to expose logical error, not to argue
about what the real world is like, I don't think the logical
argumentation is embellishment. In other words, my critique in such
cases is not that the underlying model is wrong, bad, inapplicable to the
real world, etc., but that the claims of *necessity* or *impossibility*
that have been made on its basis do not hold water.
Sometimes I make a point that Marx's profit rate is measured on the basis
of the actual capital advanced, not the replacement cost of capital.
When that is my point, I argue not on the basis of logic, but on the
basis of the textual evidence, including the ability of that measure of
profitability to yield his theoretical results.
Sometimes I make a point that firms care about and base their decisions
on the rate of return measured on the basis of the actual capital
advanced, not the replacement cost of capital. When that is my point, I
again argue on grounds other than logic. I refer to the measures of
profitability that firms actually use, such as internal rate of return
and present value, which do not constrain input prices to equal output
prices. I also argue in terms of their self-interest. If computer
prices fall in half year after year, a firm deciding whether to invest in
1 computer that will produce 2 computers would have to be extremely
myopic, and dumb, to go ahead with the investment, even though the
"replacement cost profit rate" is (2 - 1)/1 = 100%. Its actual rate of
return would be ([1/2]*2 - 1)/1 = 0%.
Andrew Kliman
This archive was generated by hypermail 2b29 : Sat Sep 30 2000 - 00:00:05 EDT