Review of Kindleberger's "Manias, Panics and Crashes: A History of Financial Crises"

_{[Note for bibliographic reference: Melberg, Hans O. (1998), Not bad, but more
popular than it deserves: A Review of "Manias, Panics and Crashes: A History of
Financial Crisis", www.oocities.org/hmelberg/papers/981006.htm]

Not bad, but more popular than it deserves
A Review of "Manias, Panics and Crashes: A History of Financial Crises"

_{Manias, Panics, and Crashes

A History of Financial Crises}

Charles P. Kindleberger

John Wiley & Sons, New York, 1996 (third edition, first in 1978)

263 pages, ISBN: 0 471 16171 3 (paperback)

Introduction

To make The Beatles popular their manager bough enough records to
get them into the hit lists. People then bough the record simply because the group
appeared to be popular. But, by so doing the group became genuinely popular; popularity
was self-fulfilling and self-sustaining. To some extent the same goes for academic works.
I read Charles P. Kindleberger's book Manias, Panics, and Crashes: A History of
Financial Crises because so many articles on economic collapse referred to
Kindleberger. It turned out that the popularity was at least partially misleading.
Although the book has its good sides, there are also many flaws.

In this review I shall first present Kindleberger's argument and the questions discussed
in the book. I shall then criticize the book under three headings: the choice of issues to
be discussed, the method employed to prove points and the conclusion from the arguments.
This negative criticism will then be balanced by a discussion of some of the virtues of
the book.

His argument

"The heart of this book", writes Kindleberger, is the view that
Keynesian and Monetarist theory are incomplete because they leave out "the
instability of expectations, speculation, and credit and the role of leveraged speculation
in various assets" (p. 18). This instability leads him to argue that "our
conclusion is that money supply should be fixed over the long run but be elastic during
the short-run crisis. A lender of last resort should exist, but his presence should be
doubted" (p. 9).

When asked why the system is unstable, Kindleberger's answer is that "What happens,
basically, is that some event changes the economic outlook. New opportunities for profits
are seized, and overdone , in ways so closely resembling irrationality as to
constitute a mania" (p. 2, my emphasis). In other words, psychological forces
stimulate speculation (causa remota) and the crash comes when some incident snaps
confidence (causa proxima) (p. 92).

There are, of course, lots of blanks to be filled in this argument. We need to know the
various links and mechanisms that create speculation, asset price increases, the links
between the asset price collapse and the real economy and so on. Moreover, the explanation
should be compatible with the fact that crashes do not occur permanently; usually the
economy works fine.

A more detailed outline of his theory, which he claim is build on the Minsky model, is as
follows. First, there is some exogenous shock (policy switching, technology, financial
innovation). Second, the boom created by the profit opportunities after the shock is fed
by increasing money supply. There is little to be done about this because the money supply
is endogenous (new banks enter, personal credit increases, new credit instruments are
used). Third, the boom leads to speculation that initially has positive feedback;
speculators earn money and invest more as well as making more people invest. This leads to
what Adam Smith calls "overtrading" which can be caused by pure speculation
(buying something with the aim of selling it later at a higher price), overestimation of
the true expected return and excessive gearing (low initial cash requirements when buying
something). Fourth, the overtrading spreads from one market to another (psychological
links and others). Fifth, speculation spreads internationally (again, psychological
mechanisms are important, as well as: arbitrage, foreign trade multiplier, and capital
flows). Sixth, at the peak some insiders leave the market, there is "financial
distress" and a bankruptcy or the revelation of a swindle leads to the final stage;
the rush for liquidity. The panic (or "revulsion") feeds itself until prices
become so low that people are tempted to once again go into less liquid assets, trade is
cut off or a lender of last resort convinces the market that three is enough money for
all.

In sum, economic crises are caused by psychological mechanisms which lead asset prices to
be volatile. Combine this with a bank and credit system that is fragile. When asked why it
is fragile, Kindleberger answers: "The system is one of positive feedback. A fall in
prices reduces the value of collateral and induces banks to call loans or refuse new ones,
causing mercantile houses to sell commodities, households to sell securities, industry to
postpone borrowing, and prices to fall still further. Further decline in collateral leads
to more liquidation. If firms fail bank loans go bad, and then banks fail. As banks fail
depositors withdraw their money (...) Deposit withdrawals require more loans to be called
..." (and so on , p. 97).

Finally, the link between the financial and the real sector is not primarily
that people buy less because falling asset prices leads to less wealth, but that the
credit sector is "paralyzed." Thus, psychological forces are magnified through
the banking system which, in turn, are magnified in the real sector. The story is one of
feedback (non-linearities), discontinuities and differences in adjustment speed.

The issues

Stylistically I think the book could have benefited from the inclusion of a short
overview of financial crisis. For readers unacquainted with the history of financial
crisis it becomes confusing when the author writes that the crisis of year x, y, and z
prove a point. True, there is a good table in appendix A with an overview of crises, but a
short verbal history would have been better.

Second, I sometimes found the argument slightly unfocused. For instance, the argument does
not require him to discuss whether some cultures have a greater tendency to speculate than
others (p. 41-43) or the appropriate punishment of speculators (p. 81-82). True, these are
minor flaws and on the whole his discussion sticks to the real argument.

A third missing element, was the relative lack of analytical discussion of the precise
mechanisms behind the crisis. Consider, for example, the argument that speculation is
cause by (irrational) psychological mechanisms. While this may well be true, we want a
close description of the precise mechanisms. Maybe people have a tendency to ignore
negative information, maybe there is herd-behaviour, maybe making money makes you
overconfident in your own abilities (as suggested by Galbraith). These, and other
mechanisms, must be discussed more closely if we want to explain crisis. Once again,
Kindleberger touches the issue (see p. 23), but the discussion is too brief given the
importance he gives speculation as a cause of crisis.

Finally, I think a closer discussion of the definition of crisis is required. He claims
that it may be hard to define but recognizable when encountered (p. 3). But I do not find
this satisfactory. Later he claims that " ... we are concerned primarily with
international crises involving a number of critical elements - speculation, monetary
expansion, a rise in the price of assets followed by a sharp fall, and a rush into money.
Crises that fall outside these dimensions do not, on the whole, concern us, and there are
enough within the category to suggest that the broad genus is worthy of study" (p.
17). Yet I am unsure whether this is precise enough, and more seriously I wonder whether
it represents arbitrary domain restriction (see below for more on this).

The method

In his own words, the book is "literary economics" as opposed to
mathematical economics (models) and statistical economics. Moreover, his approach is not
guided by the assumption that people are always rational. Finally, the evidence he uses is
largely historical correlation. For instance, to examine whether a lender of a last resort
is good or a bad, he simply contrasts the outcome of crisis with or without a lender of
last resort.

On the first - literary economics - I have no serious complaints. It is fully possible to
write good economic without using mathematics, but I also tend to believe
that formalization tend to make our understanding more precise and may reveal potential
inconsistencies.

I also agree that we cannot by assumption exclude irrational behaviour when
we explain an empirical phenomenon. However, the way to prove the irrational causes is not
to list the words used by contemporaries to describe speculation (craze, rush, mad,
feverish, blind passion and so on, see p. 22). The point is that what might seem crazy,
may turn out to be rational once we go deeper into the issue. For instance, ethnic
violence used to be explained by irrational and ancient hatred, but recently
"rational" models have increased understanding of some of the mechanisms
involved (tipping models, the impossibility of credible commitments, the benefits of
pre-emptive strikes). Hence, phenomena which appear irrational, may be explained by
rational individual behaviour. It seems to me an important task of the social sciences to
investigate these possibly counterintuitive truths and not to accept the judgment of
commentators at face value.

Finally, and most seriously, there is a problem with the way Kindleberger uses history to
"prove" his argument. Consider the mentioned argument that a lender of last
resort is beneficial. The argument is as follows: "... we are prepared to make the
case, tentatively, that a lender of last resort does shorten the business depression that
follows financial crises. The evidence turns mainly on 1720, 1873, 1882 in France, 1890,
1921, and 1929. In none of these was a lender of last resort effectively present. The
depression that followed them were much longer and deeper than others ..." (p. 190).

While it would be absurd to deny evidence of this kind, the conclusion becomes much more
convincing when it can be made using a model where the mechanisms that show why a
lender of last resort is good. As always Kindleberger has sensible suggestions, but the
impression is unavoidably that we are dealing with ad hoc reasoning. The
reason I stress this, is that using historical correlation to prove an argument can often
be very deceptive because it may be spurious and accidental (see my paper "Against
Correlation").

Only when you have a general frame and a detailed account of the mechanisms involved, can
you claim to have a reliable explanation. Reliable because you reduce the problem of
spurious correlation; Explanation because you make the causal links explicit, unlike a
mere correlation.

Finally, Kindleberger explicitly excludes small economic fluctuations from his study. As
he writes: "We are not interested in the business cycle as such, ... but only in the
financial crisis that is the culmination of a period of expansion and leads to a
downturn" (p. 1, see also p. 17 for a limitation of the subject). Initially this may
seem a perfectly acceptable way of limiting a large subject. However, when he claims that
the story should be judged by its ability to explain crisis, he tries to shield himself
from the argument that his theory is contradicted by the many calm periods of capitalism.
In short, if capitalism is as unstable as Kindleberger argues, why do we not experience
more crisis and less calm? I think this is a valid argument against Kindleberger and he
should not be allowed to side-step the issu e by arguing that he does not want to explain
the whole business cycle. (Note: While I believe in the argument made above, I have to
admit that I am less than perfectly certain. It is legitimate to restrict the domain of
your research. The question is what restrictions are valid. See The Pathologies of
Rational Choice Theory by Green and Shapiro for more on arbitrary domain
restrictions.)

The conclusions

The main analytic conclusion is that irrational speculation leads to excessive asset
prices which collapses and causes real problems because of the fragility and importance of
the banking system. The main policy conclusion "is that money supply should be fixed
over the long run but be elastic during the short-run crisis. A lender of last resort
should exist, but his presence should be doubted" (p. 9). Both conclusions can be
critizised.

One the first issue, Kindleberger does not really go into alternative explanations of
economic fluctuations and crisis. One such alternative is that the economic system is
pervaded by non-linearities and discontinuities. In such a world asset prices may
fluctuate widely even if people are close to or perfectly rational. In other words,
instead of blaming irrational people, one might blame the instability of the technology of
the system. This leads to chaos and catastrophe models that Kindleberger admits is beyond
his competence.

On the second issue he admits that "The dilemma, of course, is that if markets know
in advance that help is forthcoming under generous dispensations, they break down more
frequently and function less effectively" (p. 4). Yet, this admission is not followed
up by truing to decide the issue empirically or analytically. Even if there have been
fewer banking crisis after the Federal Deposit Insurance Corporation was introduced, does
not prove the point since many other variables have also changed since then.

This is not to say that I disagree with Kindleberger, only that in a perfect textbook on
economic crisis I would like to have a more general framework with alternative theories
and policy implications compared. I want more detailed mechanisms and more analytic
arguments within an encompassing logical structure. This is, of course, asking for too
much, but it is valid to argue that a general book should provide more of this than
Kindleberger does.

Praise

It follows from what I have said so far that I seek the details of the mechanisms that
produce crisis. Sometimes Kindleberger gives the reader new and possibly important
mechanisms and for this he deserves credit. Here are some examples.

Consider first a type of mechanisms that fits Jon Elsters definition, that is - a
frequently occurring and easily recognizable causal pattern that are triggered under
generally unknown conditions or with indeterminate consequences (p. 45 in Hedstr�m and
Swedberg: Social mechanisms ). Here are three quotations which illustrate
this type of mechanism:

"Tight money in a given financial center can serve either to attract funds or to
repel them, depending upon the expectations that a rise in interest rates generates."
8

"A rise in the price of a commodity may lead consumers to postpone purchases in
anticipation of the decline, or to speed them up against future increases." 8

"Where foreigners took raising of the discount rate as a sign not of strength but of
weakness, and sold rather than bought the currency ...." 165

In short, the same action (tightening monetary or fiscal policy, raising the interest
rate, increasing prices) can have different effects and it may be impossible to determine
the effect in advance. Yet, after the fact we see which effect that was the strongest in
aggregate and can use the mechanism to explain what happened; We may explain but not
predict.

Another common fallacy, is the fallacy of compositions. As Kindleberger writes this is
when The action of each individual is rational - or would be, were it not for the fact
that others are behaving in the same way (p. 28). This may lead to cycles by way of
overinvestment (and resulting underinvestment and possible cob-wed cycles). For instance,
observing that railroad investments are currently very profitable, many people may invest
in railroads and the aggregate effect is that railroads become a very unprofitable
business (overinvestment). This may help explain the overinvestment in railroads in the US
in 1840-50. It may also explain some of the cycles in the Norwegian shipping industry.

A third mechanism is the kind of positive feedback already quoted in this review
("The system is one of positive feedback. A fall in prices reduces the value of
collateral and induces banks to call loans or refuse new ones, causing mercantile houses
to sell commodities, households to sell securities, industry to postpone borrowing, and
prices to fall still further p. 97).

In addition tot he mechanisms suggested by Kindleberger, I found that parts of the book
was entertaining and well written. It is also a veritable database of small anecdotes,
quotes and facts. It is useful to know that during the condominium craze in Boston in 1985
and 1986 about 60% of the people buying did so with the objective of reselling (p. 26).
Another interesting fact is the use of postdated checks from 1977-82 on the Kuwait Souk
al-Manakh (stock exchange). Unsurprisingly, it turned out that some of these cheques were
worthless (totaling $91 billion! See p. 45). The incident should illustrate that with
enough ingenuity (and dishonesty) the market will find ways to increase the money supply.

Conclusion

The mechanisms that led me to buy this book may help to explain the subject of the book; I
bought it because it was popular. By so doing I contributed to its popularity and further
sales based on existing popularity. In other words, there is a self-reinforcing
mechanisms. Cycles and crashes may arise by similar mechanisms in many levels and sectors;
I buy an asset because other buy and by do doing I make the price increase which confirms
the wisdom of buying that asset - which in turn may lead other people to buy and so on.
Kindleberger gives us some of these mechanisms and he allows irrationality, confidence,
expectations and psychology to play a role that many economists try to avoid. For this he
deserves praise, but the book cannot be considered the final book since it does not have
an encompassing frame in which the various theories are considered (such as modern
non-linear theories). I would also like to see more mechanisms derived from a logical
structure and less ad hoc reasoning. In short, unlike The Beatles
the popularity was only partially a sign of good quality.

[Note for bibliographic reference: Melberg, Hans O. (1998), Not bad, but more popular than
it deserves: A Review of "Manias, Panics and Crashes: A History of Financial
Crises", www.oocities.org/hmelberg/papers/981006.htm]}