Here's a statement of the obvious: The opinions expressed here are those of the participants, not those of the Mutual Fund Observer. We cannot vouch for the accuracy or appropriateness of any of it, though we do encourage civility and good humor.
Found this Rick Feri article from Jan 2014 while surfing. He makes the case that it is harder to predict correlation between two asset classes based on past data as it can change.
I really like Rick Ferri, both as a financial advisor and as an author. I own several of his easy to read books and his advice makes investment decision making easier. I’m frequently on the same investment page as he is. Occasionally we do have a parting of the ways.
Your referenced article is one such instance, at least partially so. In this case, Ferri gets a little sloppy in his calculations.
My bifurcation with Ferri’s prospective on correlation is that he simplified a bit too much. And as Albert Einstein cautioned: “Everything Should Be Made as Simple as Possible, But Not Simpler”. I’m afraid that Ferri went a bridge too far in his article by oversimplifying the portfolio standard deviation relationship.
For a two major class portfolio, the complete equation for the portfolio’s volatility is not a simple average. It includes a corrective term for the interaction aspects of the holdings. These components typically do not perfectly zig while the other zags.
To be mathematically precise, the portfolio standard deviation (SD) squared includes an additional positive (plus in the equation) term evaluated as the product 2 X Weight of component One X Weight of component Two X SD of component One X SD of component Two X Correlation Coefficient. A portfolio’s standard deviation calculation gets more complicated as the number of components increases.
The equation presented in Ferri’s article is wrong; it is incomplete. Correlation Coefficients between investment classes are almost never Zero, and if they are temporarily Zero, they surely do not remain so for very long. Ferri does have it right when he recognizes that correlations are NOT stable in time.
But ignoring the impact of the dynamic component Correlation Coefficient’s interactions on a portfolio’s performance is a hazardous simplification. Typically, it results in an underestimation of the portfolio’s volatility because most category correlation coefficients reside in the positive domain.
Since portfolio compound return is negatively influenced by its volatility, if an investor underestimates standard deviation, he/she will overestimate compound return and cumulative return while underestimating the number of negative annual returns. None of this will inspire the portfolio’s owner confidence level. Over time, his results will fall short of expectations. It’s equivalent to ignoring the wind drag influence on an automobile’s gas mileage performance, and the car running short of fuel before the destination is reached.
I do agree with Rick Ferri’s assessment that correlation coefficients should be judiciously judged given their historical high volatility. Given the high volatility of most investment statistics, their utility in decision making must be tempered by a recognition of that volatility. Uncertainty always exists in the marketplace, so the stats merely give some guidance as to what might happen, but will never yield the precise answer as to what will really happen.
If you are interested in category and/or individual mutual fund/ETF correlation coefficients you might be inclined to visit the Linked Portfolio Visualizer website that provides a comprehensive and easy to use tool. It facilitates the calculation of correlation coefficients between a host of investment products for whatever timeframes suit your purposes:
I hope this helps. Please enjoy playing what-if scenarios with the referenced website. Correlation Coefficients are not constant, but nothing else is so in the investment world either. Good luck in your investing adventures.
Comments
I really like Rick Ferri, both as a financial advisor and as an author. I own several of his easy to read books and his advice makes investment decision making easier. I’m frequently on the same investment page as he is. Occasionally we do have a parting of the ways.
Your referenced article is one such instance, at least partially so. In this case, Ferri gets a little sloppy in his calculations.
My bifurcation with Ferri’s prospective on correlation is that he simplified a bit too much. And as Albert Einstein cautioned: “Everything Should Be Made as Simple as Possible, But Not Simpler”. I’m afraid that Ferri went a bridge too far in his article by oversimplifying the portfolio standard deviation relationship.
For a two major class portfolio, the complete equation for the portfolio’s volatility is not a simple average. It includes a corrective term for the interaction aspects of the holdings. These components typically do not perfectly zig while the other zags.
To be mathematically precise, the portfolio standard deviation (SD) squared includes an additional positive (plus in the equation) term evaluated as the product 2 X Weight of component One X Weight of component Two X SD of component One X SD of component Two X Correlation Coefficient. A portfolio’s standard deviation calculation gets more complicated as the number of components increases.
The equation presented in Ferri’s article is wrong; it is incomplete. Correlation Coefficients between investment classes are almost never Zero, and if they are temporarily Zero, they surely do not remain so for very long. Ferri does have it right when he recognizes that correlations are NOT stable in time.
But ignoring the impact of the dynamic component Correlation Coefficient’s interactions on a portfolio’s performance is a hazardous simplification. Typically, it results in an underestimation of the portfolio’s volatility because most category correlation coefficients reside in the positive domain.
Since portfolio compound return is negatively influenced by its volatility, if an investor underestimates standard deviation, he/she will overestimate compound return and cumulative return while underestimating the number of negative annual returns. None of this will inspire the portfolio’s owner confidence level. Over time, his results will fall short of expectations. It’s equivalent to ignoring the wind drag influence on an automobile’s gas mileage performance, and the car running short of fuel before the destination is reached.
I do agree with Rick Ferri’s assessment that correlation coefficients should be judiciously judged given their historical high volatility. Given the high volatility of most investment statistics, their utility in decision making must be tempered by a recognition of that volatility. Uncertainty always exists in the marketplace, so the stats merely give some guidance as to what might happen, but will never yield the precise answer as to what will really happen.
If you are interested in category and/or individual mutual fund/ETF correlation coefficients you might be inclined to visit the Linked Portfolio Visualizer website that provides a comprehensive and easy to use tool. It facilitates the calculation of correlation coefficients between a host of investment products for whatever timeframes suit your purposes:
http://portfoliovisualizer.com/asset-correlations
I hope this helps. Please enjoy playing what-if scenarios with the referenced website. Correlation Coefficients are not constant, but nothing else is so in the investment world either. Good luck in your investing adventures.
Best Regards.