After discussing the absurdity of the capital-rich United States being by far the world’s biggest borrower and biggest debtor, Brad de Long observes
it is a Dark Night of the Soul for us neoclassical economists who believe in the long run and in the even partial rationality of international capital flows.
I went through the necessary loss of faith a few years ago. As I observed last October
If you accept that the $US has to depreciate at some time, then holding bonds denominated in $US, and receiving interest rates lower than those obtainable in other currencies, is a dumb idea. Unless you think either that European governments are likely to default on their debt or that euroland is poised for inflation, eurobonds are a better bet, and similarly for Australian government bonds denominated in $A. But I’ve given up even the residual belief in the efficient markets hypothesis that would lead me to try and work out a coherent explanation of perverse asset prices. (Emphasis added)
Despite the stunning contrary evidence of the past five years, residual belief in the idea of efficient capital markets continues to shape the thinking of most economists (including me, when I let myself be guided by instinct rather than careful reasoning). Only when we have freed ourselves of this incubus will we able to make progress in understanding the global economy. Because this is akin to a loss of religious faith it will be a difficult and painful process.
John Quiggin 
There’s also the problem of something empirical, coherent, and testable to replace it with…
Jim:
I don’t think you meant it that way, but you’re inviting comments about drunks and streetlights…
In all honesty, the EMH itself doesn’t score particularly well on coherence (cf. no-trade theorems) or empirical testability (Roll critique)
I have nothing interesting to say about this topic, so I’ll go off on a tangent. Benoit Mandelbrot has a video lecture at MITWORLD under the following abstract:
“Roughness is ubiquitous and a major sensory input of Man. The first step to measure and simulate it was provided by fractal geometry. Illustrative examples will be drawn from the sciences, engineering (the internet) and (more extensively) the variation of financial prices.” (emph. added)
He does mention Louis Bachelier in passing, whose work was of seminal importance for the random walk theory, if not the EMH.
Well I found it interesting…
err, and the URL is here:
http://mitworld.mit.edu/video/52/#
financial markets are to capitalists as political markets are too statists – source of rent-seeking income distributive gains.
The traditional theory of the pluralist democratic state assumed symmetrical information sets and cheap exit/effective voice.
But this theory is no longer accepted in the light of evidence about parties and bureaucracies.
The traditional theory of the finance market firm will also have to be revised to take account of the pervasive opportunities for insider-trading, management rorting and rational short-term exuberance (scams).
The EMH is quite testable. That’s the point, it has failed the tests.
Of course that is the Long term version of the EMH. You still can’t predict it short term. If you could, you would be very, very rich.
Bond traders move in mysterious ways.
I think the EMH is more the preserve of economists rather then anyone that is actually in the market.
One more indication that rationality at best partially guides human beings. Conclusion for economists: throw the assumption of rational market participants over board once and for all.
>>The EMH is quite testable. That’s the point, it has failed the tests.
It’s not testable, hence my offhand reference to the Roll critique (which is actually a critique of the CAPM).
Any “test” of the EMH is always a joint test of EMH plus some hypothesis about the underlying process generating stock returns. EMH can always be retained if you change your assumption about the underlying process. Thus, small-firm and price-book anomalies have regularly been assumed away by suggesting that they proxy for unobserved risk factors.
In general, though, *no* theory about unprepeatable historical events is testable except under very restrictive assumptions, and EMH is a theory of this sort.
I would really encourage people to check out market microstructure theory.
Whereas standard micro theory concerns the analysis of equilibria, MMT explores the processes by which these equilibria are reached – that is, it models the price discovery process in financial markets. Agents can still be perfectly rational, but because of their different objectives (market-maker, speculator, once-off investor) and asymmetries in their information sets, you can get seemingly crazy market movements.
A very good book is that of Richard Lyons on the microstructure of the FX market – check out this link to info on his book including the text of the first chapter, or his home page for further links.
I’ve mentioned this field of economics/finance on this blog before, but I think it deserves to be more widely known than it is at present. I think it has implications for microeconomics in general – rather than purely financial markets.
Mark, this looks interesting – I read the sample chapter with interest. The combination of market microstructure and behavioral finance models seems likely to achieve improved realism, but not to have any of the policy implications associated with EMH. Anyway, you’ve convinced me this is something worth looking into.