Doug Henwood pointed to Robert Shiller's work on financial market bubbles::
> the existence of bubbles has more adherents now than in the heyday
> of efficient market theory in the 70s and 80s, thanks in no small
> part to Robert Shiller's work. For a stock market example, see
> <http://www.econ.yale.edu/~shiller/online/jpmalt.pdf>, exhibit 4.
>
> The US stock market bubble of the 1990s reverted to mean, but never
> really went beyond that. So there's reason to believe the fat lady
> is still just warming up.
>
Before Shiller, Kindleberger, basing himself on Hyman Minsky's model of financial market instability, had also pointed to irrationality as a perennial psychological feature of bubbles (see Kindleberger's Manias, Panics and Crashes).
Shiller, Kindleberger and Minsky all fail to notice, however, that Keynes had already made irrational psychology the key to understanding financial market functioning in general and bubbles in particular. Keynes understood this irrationality in terms of psychoanalytic (rather than, as in the case of Shiller, "behavioural") psychology.
Another key feature of Keynes's approach that differentiates it from the approaches of Shiller, Kindleberger and Minsky is the presumed inconstancy of the psychology involved. Keynes adopts an 'internal relations" view of social phenomena. On this view, the regularities observable in human action change with changes in the psychological identities of the actors brought about by changes in the customary social relations within which they develop and live. Thus Keynes points to the cross culturally and historically very different roles played by gold in liquidity preference and claims that the equity premium can't be relied on to remain a feature of financial market functioning simply because it has characterized such functioning in the past.
For Keynes, the importance of psychology and internal relations in social phenomena makes it a mistake to model social theory on natural science, particularly (as has been conventional since Adam Smith) on physics. For this reason, mathematical and statistical methods have a much more limited applicability in the former than in the latter. For example, the form taken by bubbles will vary with variations in the psychology expressed in them. Also, as Keynes long ago pointed out in criticism of Jan Tinbergen's early work in what became known as "econometrics," before valid statistical estimation of such a form can be undertaken, it must first be established that such a form exists. Statistical estimation itself cannot demonstrate this.
This is another way in which Keynes's (and Alfred Marshall's and Marx's) approach to economics differs radically from Milton Friedman's. Friedman uncritically accepted physics as the appropriate model for "science" in general. Moreover, he convinced himself that his statistical investigations of a quantity theory approach to the demand for money had uncovered a physics like regularity in social phenomena, a "uniformity ... of the same order as many of the uniformities that form the basis of the physical sciences." Keynes, however, had already pointed out, on the basis of a more rational "empiricism" than Friedman's, that the psychological nature of liquidity preference made such statistical estimation procedures inapplicable. This point is ignored in discussions of the subsequent predictive "failure" of Friedman's quantity theory.
Ted