Eggs Not Enough: The Truth About Diversification
October 22, 2012
Page 1 of 3
Simple as the idea is, most investors do not hold portfolios that are even close to being truly diversified. Two reasons make this sensible objective difficult to achieve. First, most investors are not disciplined enough to implement diversification. To illustrate my point, pause and check whether you are willing to reduce equities when the trailing 12-month return on stocks is 20+ percentage points higher than bonds?
Second, but not less importantly, most investors do not actually diversify their equity risk with their investment decisions; they are still exposed to one common significant shock. Returning to our initial definition, many portfolios look like a truck with several baskets of eggs loaded on it. Clearly, investors’ eggs are vulnerable to the truck tipping over.
This issue of Fundamentals will show why it is so important to get your estimates of risk correct in asset allocation decisions.
Source of True Diversification
My more investment-sophisticated friends own a portfolio with multiple asset classes including equities, bonds, commodities, etc. And within each category, they diversify across geographical or economic regions. For example, they hold both U.S. Treasuries and Emerging Market Sovereign Bonds. They also believe they are adequately diversified. They are better diversified than my less sophisticated friends, but they are probably not as diversified as they want to be.
The truth is, static diversification across multiple asset classes is not sufficient. An adequately diversified portfolio should also be diversified over time and over different economic regimes. Yes, the tactical element in your allocation!
In standard finance applications, asset class volatilities and correlations are usually assumed to be constant over time for simplicity. For example, Harry Markowitz’s mean-variance optimization requires that the asset class variance–covariance matrix is known and constant over the holding horizon. While this simplified assumption reduces the complexity of the models and their calculations, it could also lead to sub-optimal portfolios and risk management solutions. If equity market volatility is time-varying and is negatively correlated with equity market returns, ignoring this counter-cyclicality could lead to excess allocation to stocks when forward-looking risk for stocks is high. Furthermore, if equity market volatility is positively correlated with the volatilities of other asset classes, ignoring this correlation would again lead to an overall overconcentration in risky assets.