The efficient markets hypothesis has so few defenders these days that in a sense Roman Frydman and Michael Goldberg score an easy goal in their new book, Beyond Mechanical Markets. However, their ambition is not just to knock down a hypothesis many people already find implausible, but rather to make a deeper point about the nature of financial markets. This is that no conventional modelling approach can describe how markets behave, including the newly fashionable behavioural models.
The reason lies in our imperfect knowledge. Uncertainty about the assets being traded is profound. Nobody has any idea if a new tech product will succeed or flop, so there is no 'fundamental' share price to be captured by stockmarket trades. Investors and traders are constantly evaluating new pieces of information and trying to make sense of their implications, impossible as this is in a complex and changing world. “Who in the late 1970s could have predicted the phenomenal rise of the personal computer and the internet,” they ask (p179)?
The book does a thorough job, nevertheless, of demolishing what could be described as a hardline rational expectations approach to financial markets. This is the idea that financial markets capture the reality of the world which is thus reflected in prices, give or take random events. I enjoyed their discussion (in Chapter 4) of its logical inconsistencies. If everybody in the market knows the underlying 'truth' there could be no trades; and if you assume that traders hold different views in order for the model to have any trading, then how can they be rational? Some of them must be wrong.
I'm less convinced that Frydman and Goldberg account for the empirical evidence that fund managers never beat the market and that price anomalies are quickly arbitraged away, evidence which implies the truth of a weaker version of rational self-interest. No doubt this was me being slow, but I couldn't see how to fit that evidence with their framework.
The authors then go on to explain that formal behavioural economics models make a similar error, however. These models adopt predictable rules of thumb about investor behaviour when the point is that seriously imperfect information makes financial markets inherently unpredictable. They link this back to Keynes's famous comments on the psychology of markets, his comparison of investing in shares to predicting the popular winner of a beauty contest. Although there's no question that Keynes had some thoughtful things to say about markets, I don't think Frydman and Goldberg's line of argument needs this validation – referring back to the sayings of 'the master' (as per Robert Skidelsky's latest book Keynes: The Return of the Master) is becoming more of a political than an intellectual badge. More interesting are their references to Frank Knight's distinction between risk and 'radical uncertainty (in Risk, Uncertainty and Profit, 1921), and I wish they had made a bit more of this, and its relevance to a non-stationary world – and also of the links between their ideas and Hayek, a creative thinker about the role of information, who is too often ignored because of his kidnapping by one political tendency.
The book ends with some intriguing speculation about imperfect knowledge models and macroeconomics. Macro is, heaven knows, in a sorry state, and I hope the authors will turn their attention to it next.