Thursday, 6 November 2008

Another efficient markets hypothesis

While I was slightly distracted watching an election, Tim Harford posted a few days ago about the efficient markets hypothesis.

Although efficient markets is an appealing idea and can be proved in a world of 'homo economicus', the world has not been demonstrated convincingly to behave as the theory predicts.

A suggestion as to why this may be. The traditional hypothesis says that markets are influenced only by news, which by definition is unpredictable. The apparent phenomenon of 'sentiment' seems to contradict this - sentiment surely drives the market (in both directions) beyond whatever rational level is implied by news events.

I would suggest a modification to the hypothesis: instead of being driven just by unpredictable news, markets are driven by a sentiment which is also unpredictable.

Chaos theory gives us one plausible explanation for how this could arise; the interactions between thousands of interdependent agents who influence each other in ways too complex to be analysed. Sentiment in this sense is hard to model with classical economics because there is no effective price signal to communicate between people in the market. Sentiment does not necessarily obey self-correcting equilibrium laws and find an appropriate balance point.

And so, sentiment - even though not necessarily driven by exogenous events - is still unpredictable and so cannot be factored into market prices. Thus, the efficient markets hypothesis still has the implications for investment strategy that Tim indicates: past performance does not indicate the future; don't pick stocks; and don't time the market.

But perhaps the reason it has been hard to demonstrate is that researches are looking only at news as an influencer. If they also looked at events which are not news, but are also non-predictable within a linear equilibrium model, perhaps the truth of this hypothesis would be more clearly visible.

No comments: