Wednesday, 13 May 2009

Diversification works - most of the time

Portfolio theory has taken quite a beating over the last year. I've heard a number of people comment that the 4th quarter of 2008 demonstrated that different asset classes are no longer not correlated and so the diversification effects underpinning portfolio theory no longer hold.

Of course, that is not true. Correlation is a statistical concept - probabilistic, but not definite. As such, we don't talk about asset classes being correlated or uncorrelated - rather we talking about them being more or less (positively or negatively) correlated.

Portfolio theory is based on the fact that different assets and asset classes are not perfectly correlated. However, it does not assume that they are perfectly uncorrelated.

In practical terms that means that, whilst most of the time assets and asset classes won't move in the same direction together, sometimes they will.

And that is what happened last year - they moved together en masse in a spectacular fashion with disasterous consequences.

However, whilst this may be considered to be an extremely rare occurence, it is entirely consistent with the statistical theory. And it demonstrated Nassim Nicholas Taleb's Black Swan hypothesis - that the world is more shaped by rare and unforeseen events than by the more frequently encountered

So, I believe it is business as usual for portfolio theory, albeit after a painful reminder of what happens if you ignore the full scope of the theory on which you rely.