Our proprietary research strongly suggests that the Efficient Frontier is both outdated and broken. Nothing provides greater evidence of this fact than what has happened to passively managed "diversified" portfolios over the past few years.
Table 1: Correlations Between the S&P 500 and International Stocks in Developed and Emerging Markets | |||
Time Period |
S&P 500 |
MSCI EAFA Developed Markets |
MSCI Emerging Markets |
---|---|---|---|
1970-1981 | 1.00 | 0.57 | -0.09 |
2003-2007 | 1.00 | 0.93 | 0.98 |
*Source: Guerite Advisors, Imbriglia Proprietary Research |
Portfolios constructed using Modern Portfolio Theory methods have plummeted alongside the broad market despite statistical output that suggested they would have performed better.
The statistical foundation of Markowitz's Modern Portfolio Theory is based on the assertion that asset classes — stocks, bonds, real estate, cash, etc. — are not perfectly correlated. The power of the theory rests in the framework that when combining non-correlated asset classes an investor can achieve greater returns with less risk than when combining asset classes that move together or investing in a single asset class. One point that immediately becomes clear when thinking about this assertion is the importance of data used to construct the efficient frontier — the model is only as good as the input.
It is no secret in the investment community that asset classes are becoming more correlated. For example, during the 1970's and 1980's U.S. stocks and foreign stocks did not necessarily move in lock step; however, in the 1990's and the 2000's globalization caused disparate markets to be influenced by the same factors and the benefits of balancing one of these asset classes against another diminished. This is an obvious, but important, observation.
The historical data used in Modern Portfolio Theory has fundamentally and permanently changed thus rendering its output ineffective.
Table 1 (above) shows basic correlations between the S&P 500 and international stocks in both developed and emerging markets. This data nicely frames the discussion around correlation — even at a brief glance the message is clear — domestic stocks have been much more correlated to international stocks recently than in comparison to the 1970’s. Over the last five years, the S&P 500 went from having a mild correlation to International Developed Markets (0.57) to having a near perfect correlation to them (0.93), and went from having a negative correlation to Emerging Markets (-0.09) to an even higher correlation than to Developed Markets (0.98).
Table 2: Five Year Asset Class Correlations (2004-2008) Less Twenty-Five Year Correlation (1984-2008) | |||||||||
Small Cap Value |
Small Cap Growth |
Mid Cap |
Large Cap Value |
Large Cap Growth |
Foreign Stocks |
Investment Grade Bonds |
High Yield Bonds |
REITS | |
---|---|---|---|---|---|---|---|---|---|
Small Cap Value | 0 | ||||||||
Small Cap Growth | 0.16 | 0 | |||||||
Mid Cap | 0.11 | 0.26 | 0 | ||||||
Large Cap Value | 0.16 | 0.19 | 0.02 | 0 | |||||
Large Cap Growth | 0.31 | 0.18 | 0.19 | 0.12 | 0 | ||||
Foreign Stocks | 0.47 | 0.38 | 0.29 | 0.29 | 0.41 | 0 | |||
Investment Grade Bonds | -0.73 | -0.29 | -0.54 | -0.63 | -0.37 | -0.42 | 0 | ||
High Yield Bonds | 0.11 | 0.17 | 0.39 | 0.30 | 0.35 | 0.41 | -0.55 | 0 | |
REITs | 0.16 | 0.36 | 0.32 | 0.32 | 0.52 | 0.51 | -0.73 | 0.20 | 0 |
*Source: Imbriglia Proprietary Research |
Table 2 (above) goes into more depth around the data set chosen for this study — it indicates the differences in the five year correlation coefficients and the twenty-five year coefficients. That is, if two assets were correlated perfectly (1.00) over the last five year period and only 0.75 correlated over the last twenty-five year period, the chart would read 0.25 (1.00 minus 0.75). The positive numbers in the graph indicate by how much two assets have become more correlated over the last five years versus the last twenty-five years; the negative numbers the reverse. Notice in Table 2 that all asset classes, except Investment Grade Bonds, have become to some degree more correlated over the last five years than they have been over the last twenty-five. Given what we know about the world this should come as no surprise. What is surprising is that most investment advisors continue to use outdated historical data while attempting to optimize portfolios for clients. Table 3 (right) illustrates how differently asset classes as a whole have behaved during past bear markets. As you can see during the bear markets of 1990 and 2000-2002, asset class returns varied greatly. For example in 2000 and 2001 three asset classes achieved positive returns for the year and five were negative. Additionally, the range of returns in negative returning asset classes was quite wide in all years of the prior two bear markets. |
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Correlation of Asset ClassesJuxtapose these facts to those of the most recent 2007-2009 bear market (which primarily occurred during 2008) and the data clearly supports the theory that asset classes are behaving more alike. During the 2008 decline only one asset class (Investment Grade Bonds) achieved a positive return for the year of 5.24%. All other asset classes had negative returns and all of these asset classes returned less than -28.92%. What we can conclude from the data in Tables 1, 2 and 3 is that virtually all asset classes are becoming more correlated and that returns are not only unstable, but that they evolve radically over time. This creates a unique challenge for a money manager — strategic asset allocation can only be as effective as the manager who understands the data that goes into the model. Designing properly diversified investment portfolios has become increasingly difficult and we no longer can rely on traditional asset allocation methods. SimonDavis is on the cutting edge of progressive portfolio construction. |