Among the many beliefs that were sorely tested by the financial crisis is asset allocation, which has driven investing and portfolio management for decades. The goal of asset allocation is to minimize losses from individual investments and benefit from strong-performing market areas, as well as smooth the ups and downs, by investing across various asset classes, or types of investments.
Asset allocation is based on modern portfolio theory (MPT), the work of Nobel prize winner Harry Markowitz in the 1950’s. Prior to Markowitz, portfolios were constructed based on an assessment of each investment’s risk and reward characteristics. These collections of individual decisions often led to concentration in the same investment types, suffering broad losses when these areas faltered.
Markowitz proposed that investors should select portfolios rather than individual investments. The optimal portfolio is one which has the highest expected return for a particular level of volatility or the lowest volatility for an expected return. Optimal portfolios fall on what is called the “efficient frontier”, and it is not possible to construct portfolios which offer higher return/lower risk characteristics greater than the efficient frontier. Portfolios which fall below the efficient frontier are inefficient, meaning they produce lower returns for an accepted risk level or take higher risk for an expected return.
Consider a two-investment portfolio made up of a low-return, stable investment (cash) and a high return, high risk investment (a growth stock). MPT and the efficient frontier hold that by combining these two in various mixes, the portfolio will actually have higher returns and lower risk than the stable investment alone over time. Adding other asset classes like bonds, foreign stocks, real estate and commodities create more risk/return profiles, but reduced volatility was always a benefit.
Various studies have tried to test asset allocation by replacing portfolios’ asset selections with corresponding market indexes. A popular study by Brinson, Hood and Beebower in 1986, and followed up in 1991, determined that over 90% of the portfolios’ variance in returns was explained by allocation to various asset classes, while only 7% of the variance was due to security selection. A 2000 study by Ibbotson and Kaplan examined the 10-year returns of various balanced mutual funds and reached similar conclusions.
Central to MPT is correlation, or the extent to which assets move in the same direction. Some assets move in the same direction, but not to the same degree, while others move in opposite directions in a given market environment. (A correlation of 1 means the two assets always move in the same direction, while a correlation of -1 means they always move in opposite directions.) MPT and asset allocation rely on low correlations between assets to balance ups and downs.
The problem is that correlation changes over time, as illustrated by data compiled by Pacific Investment Management (Pimco) and Ibbotson Associates. As recently as the last eight years, commodities had a very low correlation with the S&P 500, while the broad bond market actually had a negative correlation, meaning bonds usually moved opposite from stocks. Foreign stocks’ correlation to the S&P 500 increased over the last ten years from under 0.5 to over 0.8 and real estate investment trusts had a similar increase.
Mohamed El-Erian of Pimco, as quoted in the Wall Street Journal, contends that the global economy and financial innovations have influenced the perceived relationships between asset classes. It has become much easier to invest in areas like commodities and real estate and investors are quicker to rush to quality investments such as cash and Treasury bonds in times of crisis, which creates downward pressure on most other markets. Ironically, as more investors crowd into an asset class in an attempt to diversify, their willingness to dump the asset makes it less of a diversifier.
And when the crisis hit in 2008, almost every asset class suffered drastic declines. Correlations of small stocks, foreign stocks, high yield bonds, real estate and commodities to the S&P 500 soared to 0.7 or more. For the most part, Treasuries and cash were about the only places to hide. Indeed, Ibbotson analyzed each of the 45 months since 1973 in which U.S. stocks lost more than 4.5% and determined that gold, intermediate-term Treasuries and inflation-protected Treasuries were the most likely assets to produce positive returns to offset declining stocks. (But as other assets have shown, it is difficult to predict that those relationships will continue to ring true.)
What does all this mean to the average investor? Some investment managers have resolved to be more flexible in the face of market volatility, while others are consolidating previously distinct asset classes (e.g., one large category for “stocks”, regardless of size or country). Financial innovation continues, bringing previously obscure strategies like managed futures, hedging and absolute return to the investing masses.
Many asset classes still have fundamentally unique characteristics, and it may be that the latest period of increasing correlations is an aberration. Indeed, the widespread discounting of risk leading to the crisis overlooked those characteristics and inflated market values, so the downdraft and resulting high correlations were necessary to restore normal markets. The logic of reducing risk through diversification among asset classes remains powerful, so while the terminology may change, the game will remain the same.
September 2009