Re: Uncertainty and Liquidity Preference

[Paul Davidson <pdavidson@xxxxxxx>]

At 06:18 PM 7/12/1999 -0400, you wrote:
>Chip, Paul, John, David:
>
>My question isn't about liquidity preference in general.  It's about the
>rationality of holding money as a way of dealing with the fundamental
>uncertainty of the long run future yields of capital assets, i.e. the Qs in
>Keynes's definition of the marginal efficiency of capital. Keynes claims
>these are fundamentally uncertain in the sense that they are usually
>unknowable i.e. "about these matters there is no scientific basis on which
>to form any calculable probability whatever.  We simply do not know."
>

Ted:

You really have mixed a hodge-podge of ideas together -- and also some
writings that started as an undergraduate thesis wit the new revolutionary
way of looking at things and  "a long struggle to escape from habitual
modes of thought and expression" to  a new taxonomy "a change of language"
.. (Quotes are from the Preface of the GT.)

You, like Runde, Carabelli, O'Donnell and others, want to hold Keynes to
the language of the TREATISE ON PROBABILITY and worry about "weights".  I
am trying to complete Keynes's taxonomy which certainly was still evolving
after 1936-- as he wanted to write a book on footnotes to the GT , until
his heart attack and  the war intervened.  Skidelsky, in his
"Acknowledgment" to the second volume of his biography of Keynes admits to
his Post Keynesian interpretation of Keynes (p.xi) and his differences with
O'Donnell on the need for strict consistency over decades of writing.
(Don't you think Keynes learned anything since the original undergraduate
TREATISE ON PROBABILITY--.  In my many discussions with Robert Skidelsky
over the years, Robert has often said to me that I come closer to the
meaning of Keynes's GT than any one else. No wonder Skidelsky's Post
Keynesian interpretation of Keynes was being attacked by Patinkin (see p.
xi). Don saw the revolution of the GT in Kahn's multiplier, I believe it is
in the axioms that Keynes overthrew which required him to search for a
reason why people would work (which incurred marginal disutility) to earn
income but not spend all the income they earned on goods, while, in the
classical system, goods are the only are the only things that provide
marginal utility.


>This is not uncertainty in the sense of "weight".  Weight is a measure of
>the completeness of the _relevant_ evidence on which rational calculations
>of probability are based.  In the case of these future yields, however,
>Keynes claims there isn't enough relevant evidence to allow such
>calculations to be made.

Correct. Weight was an idea that he developed as an undergraduate.  I would
hate to be held to ideas that I developed as an undergraduate and did not
advance from -- and often discard.  Wouldn't you??

>
>Some interpreters, Runde for instance, try to justify interpreting
>uncertainty in this context as "weight" by interpreting "calculable
>probability" narrowly to mean numerical probability.  This attributes to
>Keynes the idea that expectations of future yield take the form of rational
>non-numerical degrees of belief based on relevant evidence.  If this is
>what Keynes means, however, he is contradicting himself by stating it
>equivalently as "we simply do not know".  Rational non-numerical degrees of
>belief based on relevant evidence are knowledge.

Runde is just wrong!

>
>This problem is avoided by interpreting "calculable" to refer to any
>"calculus", any formal logic, which enables us to pass from relevant
>evidence to reasonable degrees of belief, numerical or non-numerical, an
>interpretation consistent with Keynes's own use of the term "calculus of
>probability". According to Keynes, the amount of evidence relevant to
>forecasting the long run Qs is so small that we can't reach "any calculable
>probability whatever", numerical or non-numerical, in this way.  "We simply
>do not know."
>

Despite the chapter on evaluating the marginal efficiency of capital  which
places the expected present value of a long-lived capital asset against its
current supply (pp.136-7, GT), Keynes did not believe that  decision makers
had much reliable evidence regarding future expected quasi-rents (when
discounted to give present values or demand price) (see pages 138-9; p.
149-50, GT).  Therefore those entrepreneurs who have ":animal spirits" (p.
161) have not actually made a rational calculation but will act as if they
had made such a calculation.

In a similar context see footnote #2 on p.24  where he argues that if one
sees an entrepreneur making a positive action on production Keynes means by
"his expectations....that expectations of proceeds which if held with
certainty would lead to the same behaviour as does the bundle of vague and
more possibilities which actually make up his state of expectation when he
reaches his decision".  In other words given animal spirits that move one
to action rather than inaction, one can ex post calculate the expectation
that would have been need to be held with certainty, in order to "justify"
the action taken.

>To me it makes no sense to treat "no scientific basis on which to form any
>calculable probability whatever" as capable of degree.  What would "more"
>and "less" mean here?

I hope my previous answer helps you here!
>
>To claim that "today is still the best predictor of tomorrow's conditions"
>where tomorrow is the long run is, it seems to me, to deny the existence of
>fundamental uncertainty in the above sense. Keynes grounds his belief in
>its existence in ontology and direct experience.  You can't refute him with
>mathematics alone.  As Keynes points out in the _Treatise on Probability_,
>however, "the more complicated and technical the preliminary statistical
>investigations become, the more prone inquirers are to mistake the
>statistical description for an inductive generalisation." (VII, p. 361)

  He
>quotes Whitehead in support of this: "There is no more common error than to
>assume that, because prolonged and accurate mathematical calculations have
>been made, the applications of the result to some fact of nature is
>absolutely certain." (Whitehead, Introduction to Mathematics, p. 27)  The
>mathematics training of the "quants" at LTCM appears to have ignored this
>point.  Keynes makes this "common error" a characteristic feature of
>conventional forecasting practices.

Keynes did not know of the axiom of ergodic stochastic process when he was
writing.  The concept was developed by the Moscow School of probability in
1935.  But his argument here is that some past statistical regularities
need not be carried forward in future data ? so that these regularities are
specific to a special time setting and not a generalization.  ((You should
read about ergodic theory ? for example. P. Billingsley ERGODIC THEORY AND
INFORMATION (1978) WHERE ON PAGE 1 Billingsley states "if the laws
governing change ... remain fixed as time passes, then ergodic theory is a
key to understanding these fluctuations" Whenever "the passage of time does
not affect the set of joint probability laws governing experiments
[outcomes], then the assumption of ergodicity permits regularities to be
perceived from what might at first sight be patternless fluctuation " (p.2,
also see 64-65).

It seems obvious to me that Keynes would overthrow the assumption of
ergodicity had he known its name (See p.16 of the GT where Keynes claims
that there is a need in economics to overthrow the equivalent of the axiom
of parallels in Euclidean geometry and work out a nonEuclidean theory.


>
>Keynes assumes conscious awareness of fundamental uncertainty provokes
>paralyzing anxiety in most decision makers.

If they have paralyzing anxiety then they will try to save in the form of
some time machine (liquid asset).  For he who hesitates is saved to make a
decision another day ? while he who commits his claims on the products of
industry and their inputs, can only reverse course at great expense and loss.

 >They escape from this anxiety
>by denying the awkward fact that provokes it. They "hide from themselves
>how little they foresee" by using mistaken and irrational methods to
>forecast the Qs.  The methods (as set out e.g. in VII, p. 152) assume that
>"the existing state of affairs will continue indefinitely, except in so far
>as we have specific reasons to expect a change" and that "calculated
>mathematical expectations" are possible (even though "philosophically
>speaking ... our existing knowledge does not provide a sufficient basis"
>for this). They also mistakenly and irrationally base expectations on "all
>sorts of considerations" that "are in no way relevant to the prospective
>yield" - ignoring the requirement that rational calculations  be based
>solely on _relevant_ evidence. The first method (which explains the effect
>of increased saving on the MEC pointed to in the first para of chap. 16 of
>the GT) is mistaken and irrational because "the future never resembles the
>past - as we well know" so that it is "contrary to all likelihood" "to
>assume ... that the future will resemble the past."  (XIV, p. 124)
>

Nevertheless if anyone is to hld any liquid asset besides money than they
face the possibility of nominal capital loss ? .  Speculative bubbles occur
when the fear of such loss recedes in one's mind and past capital gains are
"expected" to continue into the indefinite future. The bull thinks he will
know sooner than the average opinion in the market if, and when, the bubble
will burst... That permits the bull (e.g., Wall Street forecaster Abbey
Cohen) to sleep at night!!



>Since the expectations that result from these methods are not rational
>calculations based on relevant evidence, "weight" (a measure of
>completeness of the _relevant_ evidence on which rational calculations are
>based) is not an applicable concept.

I agree that the weight argument is merely an attempt to "quantify" degrees
of "animal spirits".

  Liquidity preference as a method for
>dealing with the fundamental uncertainty of prospective yield can't,
>therefore, be explained as a rational way of dealing with low "weight".
>There is in Keynes a kind of rational liquidity preference connected to the
>weight of rational expectations but this isn't it.

There is no rational expectations in Keynes at least as defined by the
majority of economists.  If you continue to use the word "rational" to mean
a different thing tan the majority, then you merely create a semantic
obfuscation.  Thus I use sensible to distinguish real world expectations
from economic theory's "rational expectations".  (Again note that Keynes
was using the word "rational" before it was tied directly to the ergodic
axiom by Lucas et al.)


>Paul claims Keynes assumes that people know the future is fundamentally
>uncertain and act rationally on the basis of this knowledge.   This, I take
>it, is what he means by his claim that Keynes assumes people are
>"sensible".

No I mean lets use words consistently.  If the system is nonergodic then
people ? at least in nonroutine decisions can not be acting rationally ?
but I do not mean to say therefore they must be acting irrationally.  Hence
the word sensible.



 This assumption underpins his interpretation of Keynes's
>account of the conventional practices used to forecast prospective yield
>and of the connection between liquidity preference and the fundamental
>uncertainty of prospective yield.

>
>In earlier posts I've pointed out, as I've just done again, that Keynes
>does not treat conventional expectations as "sensible" in Paul's sense.
>
>The well known passage I quoted in my most recent response to Paul, a
>passage in which Keynes connects the fundamental uncertainty of prospective
>yield to liquidity preference by pointing to a "conventional or
>instinctive" "feeling about money" operating at "a deeper level of our
>motivation" which enables the "possession of actual money [to] lull our
>disquietude" when the "higher, more precarious" forecasting conventions we
>have been using to "hide from ourselves how little we foresee" break down,
>seems to me to show that Keynes also doesn't treat this kind of liquidity
>preference as "sensible" in Paul's sense.

This is a nonsequitir if I ever saw one!



>
>Keynes doesn't make rational "faith" in the optimistic hypothesis the usual
>basis of business motivation.  The optimism arising from "animal spirits"
>is not rational optimism; it's based on hiding from ourselves how little we
>foresee.

Yes ? so what??

>  As early as 1922 Keynes was describing the typical business man
>as "a creature of instinct and of nature, primitive in his self-expression,
>unsophisticated in self-knowledge". (IV, p. 24)  Rational optimism fully
>faces the fact of fundamental uncertainty and is unshakeable; the usual
>business kind "overlooks this awkward fact" and "is based on so flimsy a
>foundation, it is subject to sudden and violent changes .. etc." (XIV, pp.
>114-5)  For an illustration of what Keynes means by rational "faith" in the
>optimistic hypothesis look at his use of the word "faith" in his obituary
>for Julian Bell. (X, p. 360)
>
I fail to see what Julian Bell wrote about conscientious objectors ? as
quoted by Keynes on CWK X p. 360 has to do with rationality!

Paul
Paul Davidson
Holly Chair of Excellence in Political Economy
University of Tennessee
SMC523
Knoxville, Tennessee 37996-0550
office phone# (423)974-4221
fax# (423)974-4601
home phone # (423)573-9160
email: pdavidson@xxxxxxx
http://econ.bus.utk.edu/Davidson.html