Sebastien Page provides the following response to these two letters:
We thank Kenneth R. Solow and Philip Appel for their interest in our research. We would like to take this opportunity to clarify a few points. First, our research was not produced on behalf of the CFA Institute except that it was presented at their European Investment Conference last year. Second, we do not advocate that advisors should blindly rely on historical correlations. Our research merely points out empirical observations on correlation asymmetries and suggests a new portfolio optimization method – called full-scale optimization – that controls for this feature of investment returns. Advisors must indeed formulate expectations about the future before they use portfolio optimization. These forecasts might differ from historical correlations. This concept goes back to the first article on portfolio selection written by Harry Markowitz in 1952 in which he states: the first step of portfolio selection begins with judgment and experience and ends with expectations about the future; the second step starts with expectation about the future and ends with the choice of a portfolio. Kenneth R. Sollow and Philip Appel provide an argument why correlation asymmetries might persist into the future. When assets valuation rise over a long period of time they all become overvalued. At that time it becomes evident that they will fall together when risk aversion increases and de-leveraging takes place.
Our research shows that as during a plague outbreak, in times of crises assets fall together rapidly and unexpectedly; while in normal times they diversify each other a lot more. Explanations for why this empirical observation is to be repeated in the future include Mr. Solow’s and Mr. Appel’s valuation build-up argument. Investor behavior also plays an important role: in financial markets fear is more contagious than optimism. Liquidity also matters. In their thirst for liquidity, investors will sell assets that normally would not be correlated. Other explanations include the proliferation of portfolio insurance strategies; quantitative equity investors using the same Barra factors; the short-put-option positions embedded in many investment strategies and credit-linked assets that become in-the-money simultaneously; hedge funds chasing the same sources of alpha; and the list goes on. For all these reasons, the full sample correlation coefficient belies and asset’s diversification properties in markets when it is most needed. Hence the myth of diversification. We agree with Mr. Appel’s observation that people using mean-variance optimization without thinking about the inputs get the wrong answer. In fact that’s the point of our research. The full sample correlation coefficient is the wrong input because it over-estimates diversification’s ability to reduce exposure to loss. We thank Mr. Solow and Mr. Appel for giving us the opportunity to address this important question.
"No longer were there individual destinies; only a collective destiny, made of plague and emotions shared by all." –Albert Camus, “The Plague”
Sebastien Page, CFA
Senior Managing Director
Portfolio and Risk Management Group
State Street Associates
Cambridge, MA
Display article as PDF for printing.
Would you like to send this article to a friend?
Remember, if you have a question or comment, send it to .
