At 02:35 PM 8/5/02 -0400, you
wrote:
Paul,
I think there are at least two parts to your message. In
part I where you discuss the EMH, I am pretty sure we do not disagree
(though we seem to be talking past each other) on either its meaning,
implications, or inapplicability. Just to be clear, I said that the EMH
states that prices for financial assets must accurately reflect all known
information, and only unanticipated information will change stock market
prices.
But the concept of "information" means what? Information
about the future realized values will be !! But if the stream of
returns on real capital stetches into the nonergodic future, then it is
logically inconsistent to say that "information" about these future
quasi-rents (earnings?) exist today.
When people use price/earnings
ratios they are using past statistical data and assuming that the future
will reflect the statistical shadow of the past-- i.e., that the future
earnings path is ergodic. It is not talking past each other Cliff, ir is about
being logically consistent. When you suggest that the efficient market
hypothesis has some shred of real world relevance, you are saying that the
future earnings of real capital goods represented by titles to the enterprise
is ergodically determined. There is no getting away from this --even if a
Nobel Prize winer says so!
[Clifford
Poirot]
( Paul, I will again repeat one of your favorite
sayings: I have never beaten my wife and I have never said that EMH has some
shred of real world relevance. First, let us define the efficient market
hypothesis. These following excerpts from Sharpe, Alexander and Bailey, a
standard investments text, entitled "Investments". From pg.
9:
"It will later
be seen that this apparent randomness in security returns is a characteristic
of an efficient market: that is, a market in which security prices reflect
information fully."
Or, to quote
from the glossary: "Efficient Market: A market for securities in which every
security's price equals its investment value **at all times** (my emphasis),
implying that a specified set of information is **fully and
immediately**(again, my emphasis) reflected in market
prices."
The textbook authors distinguish between the strong
(information includes all private and public information), the semi-strong
(all publicly available information) and the weak (previous prices of
securities) and then offer an equivalent definition (pg. 93)
:
"A market is efficient with respect to a particular
set of information if it is impossible to make abnormal profits (other than by
chance) by using this set of information to formulate buying and selling
decisions."
If you insist on
alleging that I advocate any form of the EMH, in its strong, semi-strong or
weak form, then you must show where I have either said this, or how it follows
clearly and consistently from something else I have
said.
I will point out, that IMO, the authors of the above
cited text neglected what I see as an important corollary to even the weak
form of the EMH: that future stock prices will respond to future information
in the same fashion as it has in the past. Or, looked at another way, I
believe EMH requires that the historical beta is known and is the same as the
true beta, and that the future beta is the same as the historical and true
beta.
Let us define information and see clearly why
1) asymmetric information undermines even the weak EMH and 2)
why non-ergodicity undermines it but does not change the fact that people
respond to information.
1) Information is simply a set of data about current
stock prices, earnings, and a set of factors believed by investors to effect
earnings. Acquiring this information is costly, different people have
different levels of access to the information, people use and interpret
information differently, and there is simply too much information to fully and
completely determine the true beta. That means, any investor who uses beta as
a guide to investing is using, by definition, imperfect information. Using
past data to predict future performance **MAY** provide a decent rule of thumb
by which to make decisions-however-you simply cannot know that your
information is correct. Thus prices will and can change in response to even
anticipated information. As a corollary (leaving aside the problem of insider
information) it is at least theoretically possible for companies or
individuals, armed with better information tools to outperform the market for
periods of time.
Since this post will already be very long I will cut
to the chase and note that this requires a more extended discussion of how
people process and use information-but it amounts to **BOUNDED RATIONALITY**
and **NOT** RATIONAL EXPECTATIONS. People will make consistent errors and the
weak EMH may **appear* to hold for periods of time by accident. I will leave
it to the mathemeticians to argue if I am correct or not in my intuition that
the above is sufficient to be unable to generate a reliable probability
distribution.
2) The problem of ergodicity now relates to future
performance and guesses about whether or not future performance will have the
same parameters as the past. 1 above relates to information about the present
and past. Of course I cannot have information about the future, but I can make
educated guesses that are not true in a probabilistic sense. I merely derive
standard rules of thumb and hope for the best. But I have no way of knowing
how numerous changes and events will effect the parameters that shaped the
past.
Rules of thumb will prove wrong, but they will be
better than random guesses or wild guesses.
)
You
elaborate on this theory and note that if it is true, then stock market
prices must also accurately value real underlying asset prices. You rightly
reject this idea, as do I, and I am not sure where you ever got the idea
that I believed in the EMH.
If you do not believe in it why argue that expected future
earnings affect today's price and that changes in these expectations of future
earnings affect changes in today's price??
[Clifford
Poirot]
(Because they do. As I explain above-people are
guessing on the basis of a costly and involved process of acquiring,
processing and extrapolating on the basis of past and present information. And
information about past earnings and present earnings, and factors that
investors believe effect those earnings is my only-albeit unreliable
guide. As I think I have shown intuitively above the assumption that people
extrapolate on the basis of past information will not generate the conclusions
of the EMH even in its weak form. That does not mean that unanticipated
changes will not effect current prices-they will. But that is not enough to
get the EMH.
If I understand you correctly, you are
arguing people buy financial assets for speculative purposes only "to sell
to a greater fool" as you put it. Suppose I accept your argument for the
moment. Why would I think a financial asset would go up in value, or down in
value. In other words, what would make me want to buy and what would make me
want to sell. If I understand your past arguments correctly, you might argue
it is essentially a beauty contest with bull believing the stock beautiful
and bears seeing it as yesterday's movie starlet. To complicate matters, it
is not just what I think of the stock's "beauty" it is also what I think
others think that matters as well.
To which I
say-all well and good-but what makes people think that others think....Is it
entirely subjective?
If the future is nonergodic and
you are tlaking about the future stream of earnings then it is simply
subjective! What you might say is that people expect other people to
react to changes in (past) reported price/earnings ratios -- and therefore
they are "betting" on what people will do when information" about the
past becomes public data.
I would argue no-it
is subjective interpretations of various sorts of "objective" information
people receive that they interpret and process in numerous ways.
[Clifford
Poirot]
(As I say above-it is subjective interpreations
of various levels of objective data and how people think other
people will respond. So yes, people do bet on what other people will do. I
do not disagree with this point. I disagree that it can (or is)
arbitrary. Bounded rationality means just that-rationality that is
bounded.
The "objective information" is always about the past
-- . As I have argued innumerable times to get "objective information about
the future" requires drawing samples from the future universe-- under
controlled conditions!! Since that is impoosible (or do you deny
it is impossible?), one must invoke the ergodic axiom which states that
samples drawn from the future will possess calculated statistical averages,
variances, etc that do not vary from the statistics calculated from past
data (collected under statistical control conditions). Otherwise past
statistical averages provide NO information about future statistical
averages!! [And even Saregent has bow implicitly admitted this in his
book BOUNDED RATIONALITY IN MACROECONOMICS (1993)---]
[Clifford
Poirot]
(Point argued above at length. Bounded rationality is
better than arbitrariness but that does not mean you will be right, even in a
probabilistic sense. So I am not arguing that you can draw samples from the
future based on the past. You make guesses about future values of
parameters-but these are parameters. However, as I draw this information from
the past and construct models, I rely on a number of assumptions that drive
firm, consumer, government, institutional, etc. behavior. If I can
accurately model behavior on a large enough sample to the extent, for example,
that I can generate elasticities of demand or supply, and there are no major
changes, then it is conceivable my model will predict "as if" it is ergodic
for a period of time. In the same sense that I can predict the sun will
rise tomorrow.
Which is to argue that people look at the prospects for both
short term and long term appreciation at least in part based on the
financial health of the company (or factors that they think might affect the
financial health of the company).
No what they look out is how they expect others to react on
the basis of ACCOUNTING data [and we now know that such data are highly
arbitrary and not collected objectively under conditions of statistical
control]! And as Keynes pointed out , if you foolishly did not make
stock market bets on what you expect others to do -- rather than your
independent interpretation of the future, you would be the FOOL.
remember "conventional wisdom suggests it is better for reputation to fail
conventionallly"
[Clifford
Poirot]
(Paul, there are many assumptions made about
accounting practices, and as we have learned, it can be reported arbitrarily.
Note how the realization that the information that has been collected was
not only "imperfect" but was in fact "false". That accounting standards
incorporate an element of subjectivity does not mean you cannot
distinguish between an honest difference of applying good rules
versus outright fraud. There is a meaningful distinction to be made
between reporting the entire value of a contract (rather than the commisssion)
as "revenue" and thus inflating earnings while setting up offshore
subsidiaries and shifting your debt and choosing to depreciate by straight
line or variable line accounting methods. Note how a fairly
competent undergraduate accounting major could have, with the time,
inclination and motivation to investigate energy trading accounting practices
figured out something was "fishy" simply from company reports and SEC
filings. Why did investors not do this? Acquiring it is costly and time
consuming and they relied on the "known" information of analysts who also took
the lazy (and career promoting path) of choosing to ignore
the elephant in the
room. )
As
I pointed out to Barkley, I never have to receive a single dividend check.
All I need to know is that the company has solid earnings prospects if not
today, then I believe tomorrow. And I think it is pretty clear where I part
company with the EMH-I do not believe this information can ever be complete
and accurate or fully known in the present, and no one knows for sure what
is going to happen tomorrow.
can one be probabilistically sure?
[Clifford
Poirot]
(No-one cannot
be probabilistically sure)
But
I can make reasonable distinctions between finanacial assets that I have
reason to believe will be stable in long run returns and those that are
speculative, and those that are mere Ponzi schemes.
Well you are clearly more perceptive than mere
mortals!!---)
[Clifford
Poirot]
(No, I merely think that Minsky was right when he
distinguished between hedge, speculative and Ponzi finance. I think my auto
dealer, bank, whatever is being "prudent" when it distinguishes
between the borrower with a solid repayment history and a steady
employment history and the borrower with a poor payment history and a spotty
employment history. That does not change the fact that the loan on my new
Accord is premised on the fact that the budget for Ohio will not fall out of
the bottom and force the University to purge itself of its sole
economist. So there is an unpredictable chance here that the gamble
of my continued employment is a bad gamble-but it is a
better gamble than betting on the guy who had his lost auto
repossessed.
Similarly, there is a difference between financing
workable and feasible capital investment projects that have a realistic chance
of providing future income streams vs. funding white elephants. To wit:
Bulgaria borrowed very heavily in the late 1980's to finance imports. Had it
used the borrowing to begin overhauling chemical companies that capital
plant sizes 3 times the international average, it would be in a much
better situation to repay the loans today.
This is the history of the debt crisis-loans made to
fund white elephants and bailouts for creditors. Mobutu Sese
Seko was a clear bad gamble, even in a non-ergodic
world! Projects to fund wells for villages to irrigate is a good gamble,
even in a non-ergodic world.
As to the second part of your argument: I
have never actually seen where Stiglitz makes the "noise trader as fool
argument" though I can see why he might make that argument. I do not agree
with it and that is not what I am saying
See
Stiglitz (1989) "Using Tax Policy to Curb Speculative Short Term Spending"
Journal of Financial Services, 3, pp. 101-113.
. People
speculate because they want higher returns and believe they can get it. They
succeed enough that there is a fairly strong incentive to do so. Soros made
an incredible amount of money off the "noise" associated with the British
Pound. Using other people's money did not hurt him either.
Obviously to speak of noise in the market-- means to assume
that the stochastic system generating the data is ERGODIC. For that is
required to define NOISE in the statistical sense of variations around a
moving average mean via analyzing time series data. The trouble is
people use statistical terms without comprehending the implications of the
"dewsign of experiments" to which such terms apply.
[Clifford
Poirot]
(My view is that the social sciences are behavioral
sciences and thus cannot ever generate the degree of hardness of the natural
sciences. That does not make the social sciences arbitrary as the
post-modernists argue. It makes outcomes indeterminate-but it does allow for
rules of thumb to sometimes be
correct.)
Upon more careful thought, I might amend my
previous argument and concede people can buy postage stamps or baseball
cards for the same speculative motives people buy financial assets. I don't
think this changes the point.
Yes it
does.
They buy postage stamps because they believe the price will
rise. But why do they believe the price will rise? Stamps, as you point out,
do not give you a claim on the Post Office's assets. But as you point out,
the Post office can indeed effect the value by issuing more stamps. So
what?
Rare stamps are rare because the post office
has announced it will no longer produce them. The same is true for 1st
editions of books-- for we recognize that a xerox of a first edition is not
the same as a first edition!! The elasticity of production is zero.
That does
not change the fact that people make guesstimates about future values of
financial assets based on a subjective belief about a company's long term
financial prospects.
This is merely going around the argument again-- is it what you
believe is the future propsects or what you believe others will
believe?
[Clifford
Poirot]
(It is both. I believe others will respond
to information in a reasonable way, most of the
time.)
Growing
liquidity does not change this. The loan and mortgage consolidators buy the
loans because they see a potential for return. The sellers sell because they
would rather make new loans, and sell them, thus getting the cash today and
turning over their portfolio.
Why for every bull there must be a bear. So what? Just
think that the loan pusher banks no longer are taking on the uncertainty
associated with default-- they are exchange an uncertain return for a
contractually agreed upon fee based return! Why?
[Clifford
Poirot]
(They prefer the immediate and sure return. So
what? They also make more money by turning over the loan portfolio and making
more loans. As soon as these loan pushers perceive greater risk in the
system, they will change their rules of thumb and ration credit. Why do they
ration credit?
I
do not see how that changes the fact that people buy financial assets on the
basis of expected future value and that expected future value (though often
wrong) is based on how people interpret future prospects for
performance.
He who will not see can not be bropught to see.
[Clifford
Poirot]
(As
I point out above, the rise of securitization does
not change the fact that banks make decisions using rules of thumb based on expected
performance of loan portfolios.)
Paul
Paul Davidson
Editor, JOURNAL OF POST KEYNESIAN ECONOMICS
Economics Department - University of Tennessee
503 SMC
Knoxville, Tennessee 37996-0550
work phone: (865) 974-4221
fax: (865) 974-4601/ (865) 974-1686
home phone and fax (865) 692-0802