Re: Piorot on Madrick - Winslow's blast

[Ted Winslow <egwinslow@xxxxxxxxxx>]

John M. Legge wrote:

> Since I didn't use the term "rational maximisation" I feel no need to
> defend
> it.

From what you go on to say, I take it the problem is with the word
"maximization."     The complex system aspect makes it inappropriate to
describe rational choices as maximizing choices.

I take it, however, that you are defending the hypothesis of
"rationality" elaborated in terms of complexity theory.  It's this that
I was questioning.

> Within the environment in which human agents exist I see no problem in
> the
> argument that they make the best decisions that they can using the
> information and computing resources available to them with a view to
> progressing their objectives.

The topic of this thread, you may remember, is an NYT article by Jeff
Madrick focused on an economics that questions the assumption "that
buyers and sellers are rational and seek to maximize their own incomes
and profits," i.e. the assumption of "rational maximization."
"Behavioral economics" as represented by Akerlof is described in that
article as an attempt "to make the central assumption of economics
realistic. People are often not rational, maybe even most of the time."
It isn't just "rational maximization" that Akerlof is said to be
questioning; it's the rationality hypothesis itself.  This aspect of
behavioral economics is said by Akerlof as in "the spirit of Keynes."

> Such theories depend greatly on one assumption: that buyers
> and sellers are rational and seek to maximize their own
> incomes and profits. These economists, perhaps best
> represented by the work of the so-called new classical
> economists, whose home base is the University of Chicago,
> conclude that stocks are almost always sensibly priced.
> They also conclude that unemployment cannot be reduced
> without creating inflation, people save enough on their
> own, regulations that limit the international flow of
> capital are anathema, and even monetary policy has little
> or no consequence.
>
> Mr. Akerlof argues, however, that market bubbles can exist
> and should be kept under control; that unemployment can
> often be pushed lower by government without generating
> inflation; that people will not save enough on their own;
> and that liberalized global capital flows have been
> damaging. He argues that monetary and fiscal policies do
> matter in creating jobs and raising incomes.
>
> The starting point for Mr. Akerlof and his colleagues is to
> make the central assumption of economics realistic. People
> are often not rational, maybe even most of the time.
> Consider investors in stocks. Keynes implied long ago in
> his "General Theory of Employment, Interest and Money," Mr.
> Akerlof notes, that investors are subject to fads and
> fashions.
>
> Behavioral economists have developed a lot of evidence to
> support this idea. For example, Robert Shiller of Yale has
> carefully shown that stock prices are much more volatile
> than corporate profits and dividends.
>
> An important conclusion is that when stock prices are
> historically far out of line with profits, there is a good
> chance that it is a bubble, not a "new economy" of
> ever-higher profits, as Mr. Greenspan often suggested in
> the 1990's.

> The strength of behavioral economics is that it is based
> largely on actual observation of human behavior, not pure
> theory. "In the spirit of Keynes," Mr. Akerlof writes,
> "behavioral macroeconomists are rebuilding the
> microfoundations that were sacked by the new classical
> economists."

I understood you to be challenging this claim that the assumption of
rationality was unrealistic by calling in question Akerlof's use of
"behavioral economics" to explain a uniform wage. You suggested that, in
fact, a uniform wage "may be the best approximation to the economically
optimal level."  I understood the "economically optimal level" to mean
the level achieved when "buyers and sellers are rational and seek to
maximize their own incomes and profits."

You've previously made similar claims about the rationality of
expectations formation and liquidity preference.  In each of these cases
you elaborate rationality in terms of complexity theory.  You also
appear to claim that such explanations can be "solidly grounded in
empirical evidence."

> Recent work on expectations and behaviour cited by Mankiw of all people
> demonstrated that expectations are overwhelmingly based on the recent
> past:
> that tomorrow will be like yesterday.
>
> If you approach the economy from a complex systems viewpoint, evolving
> on a path that is incomputable without access to a system vastly larger
> than the system being studied then this is, in fact, the best possible
> guess: for 65 million years an observer of earth could have observed
> dinosaurs and hypothesized that they would still be there on the
> following day.  Such a prediction would have been 99.99999999 per cent
> accurate, which beats the average stock picker or Fed observer."

> If you use the term "fundamental" in a less fundamental way, as I do,
> and
> apply it to the problem of determining what a reasonable person might
> reasonably pay for an income-generating asset, when not influenced by
> speculative or beauty contest motives, the Dixit and Pindyck approach,
> or
> any other approach based on uncertainty increasing with the time from
> now to
> the expected event, produces a pragmatically acceptable result.
>
> Such approaches also suggest a rational liquidity preference and offer
> micro-level support to Keynes as distinct from Marshall and the finance
> textbooks which suggest that investments with an infinitesimally
> positive
> interest rate are to be preferred to holding cash.

You also identify the idea of a "complex system" with "science" and then
seem implicitly to suppose that any other way of accounting for
"uncertainty" in Keynes's sense must be treating uncertain events as
"inexplicable in hindsight" and hence must be implicitly introducing
"divine dice throwing" and permitting "science to leave."

> Non-ergodicity can arise in a fully deterministic system simply by
> linking
> events; this is why the geometric random walk is one of the simplest
> nonergodic processes. By the 1930s if not earlier economists had
> recognised
> that systems did not have to be very complicated before they lost the
> ergodic property and entered on a formally unpredictable path.  The
> motion
> of the earth and moon around the sun is nonergodic: unless an
> astronomer had
> access to infinitely precise measurements of the state of the solar
> system
> at an instant of time, and an infinitely powerful computer with which to
> process this information, she could not make reliable predictions of
> such
> ordinary events as an eclipse of the sun for more than ten thousand or
> so
> years into the future.
>
> Human society is more complex, and your example of an entrepreneur
> trying to
> predict demand for a yet-uninvented product fifteen years into the
> future is
> probably on a par with an astronomer talking about eclipses a million
> years
> or from now; but in my view there is no fundamental difference:
> unpredictability is a fundamental property of the universe.  There is no
> need to invent a source of fundamental uncertainty into a system in
> which
> fundamental unpredictability is already present.
>
> More dangerously, in my view, adding an axiom of uncertainty denies the
> assumption of causality on which science relies.  If PKers choose to
> rely on
> an axiom of uncertainty I do not know how they/we can challenge those
> of the
> orthodox school who assert fundamental ergodicity.  They implicitly
> rely on
> a God who coordinates individual choices to as to make intrinsically
> nonergodic outcomes obey ergodic laws; but PKers are asked to call
> upon a
> different God to play dice with the laws of the universe and thereby
> produce
> uncertainty in circumstances where unpredictability would have
> preserved the
> laws of physics and still been a sufficient condition to support every
> conclusion of PK analysis,

In the e-mail you're continuing to describe as a "blast," I (1)
questioned the logical coherence of the rationality hypothesis
elaborated in terms of complexity theory, (2) questioned whether the
hypothesis could be empirically grounded merely by showing that some
empirical phenomenon, e.g. a uniform wage, liquidity preference or the
basing of expectations on the recent past, could be so interpreted as to
make it consistent with the hypothesis, (3) questioned whether the
rationality hypothesis in any form could be empirically grounded given
the abundant evidence that irrationality does play a significant role in
economic behaviour, e.g. in the formation of expectations, (4)  pointed
out that Keynes's ontological hypothesis of "organic unity" - i.e.
"internal relations" - could explain "uncertainty" without requiring any
abandonment of "science" (in the sense of rationally grounded belief),
and (5) pointed out that complexity theory can't reasonably be
identified with science in this sense because it is implicitly based on
corrigible ontological premises (e.g. those required for the valid
application of any form of reasoning that makes use of the logical idea
of the "variable") and because there are good reasons, particularly in
the case of social phenomena, for believing that these premises are very
frequently not satisfied (this ontological point also underpinning the
idea of "scholasticism" as mistakenly treating what is essentially
"vague" "as if it were precise and trying to fit it into an exact
logical category").

Instead of answering any of this, you now say:

> I suppose that Ted's fundamental problem is that he sees complexity
> theory
> as a (flawed) tool for predicting the future, where it is in fact a way
> of
> quantifying the unpredictability of the future,

Not only is this not my "fundamental problem," it has nothing to do with
any of the arguments I've made.

Ted