[From Bill Powers (2004.01.07.1217 MST)]
Bill Williams 6 January 2003 10:30 PM CST –
Before I encountered control theory
I had the idea that the Giffen paradox was an anomoly that might provide
the clue as to how to over turn orthodox theory. At that point about all
I had was a notion borrowed from Veblen that value theory has to do with
“a theory of the life process” rather than just supply and
demand or financial issues. “The life process,” fairly
obviously in retrospect involves a hierachical organization of goals.
Even Veblen’s critics took note that his discussion of value sometimes
took an “alimentary” or “nutritive” aspect.
REally about all you have to do to “solve” the Giffen paradox,
or at least take a step in the direction of solving the Paradox, is
drawing the cross between the budget and the caloric line. Once the
caloric line is taken seriously, it ought to be apparent that
maximization can not provide the right answer to what a consumer must do
to maintain life when the price of the inferior or Giffen commodity
increases. That, all by itself, as a matter of logic, ought to drive the
silver spike through the heart of the orthodox
program.
I really think that the Giffen effect is just one manifestation of a much
more general result of using system analysis in economics. I think you
said something like that too. In the Giffen situation we have only a
couple of goods with two properties each and three reference levels, so
the predictions come out in the form of a relationship between prices and
consumption of goods of two kinds. But suppose we had a hundred goods and
500 prices from various suppliers, and a variety of personal preferences
and needs combined with budgetary limitations. This clearly takes the
problem beyond what any unaided human brain can comprehend. We could have
a mixture of Giffen effects and normal supply and demand, and
second-order and third-order effects which make such elementary concepts
as “marginal propensity to consume” useless. To enter into this
world, it’s essential to use quantitative simulations.
But, the orthodox program in
economic theory is much more important than a mere question of logic or
experimental results. Keynes (1936) said it better than anyone
…
I remember admiring that passage. It must have earned Keynes a lot of
hatred.
To oppose this sort of thing I
think what is needed is a largeenough body of work to create the impression of something that
amounts to an alternative worldview. One aspect of doing this
is to identify, collect and solve a large enough number of anomolies
so that rather than remaining as anomolies the cases are transformed
into exemplars that illustrate the new paradigm.
I agree that anomalies are important to find and analyze. However, when
the method of analysis is totally unfamiliar to the person who is to be
impressed, the demonstration loses a lot of its zing – the victim has to
suspect that something is being put over on him, or at least he doesn’t
understand enough to be sure it isn’t.
Not to sound my one note too often, I think that the Test Bed is a good
way around this problem. I have always felt that economists tended to
treat their subjects much as biochemists do: through a keyhole view,
which reveals relationships among adjacent variables but fails to reveal
the Big Picture, the causal loops, the detailed bookkeeping, the
conservation laws that exist, and perhaps those that are thought to exist
but don’t. My test bed idea is supposed to do what no human brain can do:
keep track of all transactions and flows of goods without ever making a
mistake or leaving a loose end to serve as a fudge factor.If you leave
out anything essential from a working model, it won’t work. If you put a
rule into a working model, it will apply even when you forget to apply it
and even when you want it to be conveniently suspended while you get past
a tight spot.
I was impressed, in the General Theory, by the way Keynes could
conjecture about complex relationships among variables. But even Keynes
could take that only so far using only his personal gray matter. In every
argument of Chapter 6, loose ends fade off into the distance where they
disappear. The big loops are never closed, so variables that actually
depend on other variables are treated as independent. As a consquence, a
lot of miracles occur. And it couldn’t be otherwise; there are just too
many variables and too many relationships among them for the human
intuition, even Keynes’ intuition, to handle correctly. Even if we
couldn’t point to a specifid mistake, we could be sure that Keynes was
mistaken. Nobody could get all those relationships right.
Marx, Marshall and Veblen all more
or less predicted that economics would eventually make use of biology.
Marx talked about treating man in terms of a“species being,” Marshall said that the economists Meca ought
to bebiology, and Veblen argued that economic theory, real economic
theory when it finally arrived, would neccesarily be based upon
“the theory of the life process.” So, in a real sense
there’s beena background in which the arrival of the new “dispensation” has
been repeatedly foretold-- and by what are some very well known
prophets.
That’s good support from the Ancients. We can do better, however, with
control theory by showing exactly what biology and psychology are
involved. Yet (beep beep) the test bed will introduce something equally
important: a correct account of the environment within which economic
transactions and models of economic behavior take place. As you point out
above, economists rely as much on belief as on observation, if not more
so. They assume that what they hope will happen must happen. The smoke
gets so thick that you can hardly see from one mirror to the next. A
simulation won’t let them get away with that. If they object that the
simulation doesn’t work correctly, we can ask them to fix what’s wrong
with it, with expert help, and then see what it will do. If they fix
everything that’s wrong and it still doesn’t behave the way they believe
it should behave, the game is afoot. Somebody is wrong. Guess
who.
While I agree that the big breakthough will be understanding “the
secret of life,” it’s going to turn out to be just as important to
get the model of the environment right. We ought to be able to do that:
it’s tedious but not complicated. And it’s actually less controversial
than the model of the economic controllers you put into that
environment.
I wrote to Ernst Mayr once about control theory (what else) but he really
wasn’t interested. In particular, I pointed out the difference between a
sequence and a program, and he simply said that his definition of a
program was good enough for him (no choice points). I’m always leery,
after lo these many years, of quotes from famous people that sound as if
they’re right on the money. Further investigation is like taking off the
rose-colored glasses.
Very nice post, Bill. Most interesting.
Best,
Bill P.