Friday, May 14, 2010

On Diversification

Pretty sure I've posted this before, but I actually have a thought about it this time. This is pretty basic to the Buffett-o-philes but there are still a lot of economists and others who see diversification as a good risk reducer.
This gets at my problem with diversification as a risk reduction mechanism. Some academic financial economists decided years ago that volatility = risk
and people have been building on that for years. Certainly, there's risk that can be diversified away, but I'd bet that if you can find 15 securities that are not identical in most respects, adding a 16th doesn't cut as much risk as it saps return.
I've never seen it framed this way, but the volatility/risk fallacy seems to me to be a correlation/causation issue. Just because going from 1 to 2 securities is less risky and less volatile doesn't mean that it's less risky because its less volatile. If the security's price fluctuates significantly, you still hold the same intrinsic security, and if nothing has changed about the underlying security, that increase in volatility is not a good measure of risk. Instead, going from 1 to 2 securities is less risky because there are fewer events that can wipe out your investment.
In other words, if you're willing to step back from the "price fundamentally represents everything publicly known about a security" (semi-strong form efficient markets hypothesis), for which there is almost no empirical support (it's a tendency which can often be true but is not always true), then going from 1 to 2 securities cuts risk because a single major fundamental change to the world reduces the intrinsic value of your portfolio by less if not all of your securities are levered to the original state of the world.
This helps explain why markets can be efficient but are not necessarily efficient. If prices successfully respond to 95% of actual fundamental changes, and also to 80% of things which seem like fundamental changes but are not, plus some noise factor, then you have a situation where if the big news around stocks are actual fundamental issues, the markets are reasonably efficient because they're pretty good at reacting to real issues. This is most likely when there is a strong diversity of opinions, though this isn't a true necessity. If the market is reacting because a large number of people think something is a fundamental change, but it isn't, then markets don't necessarily have to be efficient. This also holds if people are no longer trading on the fundamental changes but on other factors (for example, levered firms liquidating positions they like to cover margin calls, as occurred for the end of 2008). This is how you see bubbles and busts, and why it was so easy to make a fortune if you could look at the market with a degree of time frame patience  in March 2009. Regressions that seem to signal efficient markets would be improved by thinking about this mechanism.
Even people who advocate for less diversification rarely understand this or acknowledge why they're doing so, and actually DO expose themselves to more risk in the process by saying "diversification isn't a good strategy" and thus loading up in securities exposed to the same fundamental sources of risk. Diversification is not about securities, but about sources of risk that aren't always reflected in prices (with the converse holding true, that changes in price often do not reflect actual sources of risk).

No comments:

Post a Comment