A reader (Al) recently asked whether the correlation among various asset classes had risen in the last few years:
My frustration is that the Ferri data is for the period through 2001. My layman’s guess is that the impact of globalization really hit after 2001. If true, then the correlations are probably much higher since then. I don’t want to harp on this element of asset allocation but how do you find asset classes with low current (e.g. 2001-2006 correlations? Obviously, fixed assets should have a very low correlation – but the recent returns have been less than stellar.
(If the correlation coefficient (or just correlation for short) is new to you, check out my article about Elaine from the Seinfeld Show (seriously) and my series on Asset Allocation.) So I dug into some research and found some good articles published recently that may help answer Al’s question.
The short answer is that the correlation among some asset classes has risen in recent years, while others have not. More importantly, the correlation of various asset classes over the last 30 years may offer some surprises for you and change the way you invest your money. Let’s take a closer look.
The Recent Correlation Data
My data on correlation comes from two articles I highly recommend: (1) Buy the Numbers–Asset Class Correlations by Richard A. Ferri, CFA; and (2) Emphasizing Low-Correlated Assets: The Volatility of Correlation, by William J. Coaker II, CFP, CFA, CIMA. The Ferri article is a continuation of his excellent book, All About Asset Allocation, and includes data through 2006. Coaker’s piece includes first rate analysis, but the data stops at 2004. According to these two articles, the correlation between some asset classes increased, while others decreased. Here’s the data in a nutshell:
Stocks and U.S. Treasury Bonds: Correlation ranged from about 0 to -0.6. The correlation was as high as .6 in the mid-1990s.
Micro Cap and Large Cap Stocks: Correlation ranged from about .5 to .85, which was higher than in the previous decade.
U.S. and Pacific Stocks: Correlation ranged from about .45 to .6.
U.S. and European Stocks: Correlation ranged from about .7 to .9.
REITs and U.S. Stocks: Correlation ranged from about 0 to .5.
The Long Term Correlation Data
Coaker does a great job of charting the correlation data from 1970 to 2004. The averages and standard deviation of the data as compared to Ferri’s data show, among other things, that the correlation between U.S. Stocks and Developed Country stocks has risen in the last few years. Here is one of Coaker’s charts that shows the long term correlation data:
What Are The Implications for Investors?
Coaker does a nice job of describing some of the implications of this data for investors. He identifies 5 asset classes that have consistently experienced negative or low correlation to large U.S. Stocks: Natural resources, Long-short, U.S. Bonds, Global Bonds and cash. Long-short refers to funds that take both long and short positions in an effort to benefit in either a rising or falling market. What is interesting about this list is that while most investors hold U.S. Bonds, many do not own Global Bonds, natural resources or long-short funds. I own global bonds, but not natural resources or long-short, both of which I’m now researching in light of Coaker’s article.
The other interesting aspect of the article is that not all foreign fund should be treated equal. The correlation between international stocks (of developed countries) and emerging market funds averaged just .56. The point is that these two asset classes are quite distinct, and it underscores for me the importance of owning both.
Finally, as important as correlation among asset classes is to a diversified portfolio, several things must be considered:
- Correlations Change: As both Ferri and Coaker’s articles demonstrate, the correlation among asset classes changes from year to year. Making decisions based on short-term correlation data is no better than making investing decisions based on short-term returns data.
- Correlation is Just One Factor: In addition to correlation, one must examine the expected returns of an asset class. Certain commodities, for example, have negative or low correlation with the U.S. Stock market, but their returns are dismal. Ferri makes this point particularly well.
- Ignorance is Bliss: One of the justifications for diversification is that we just don’t know how various asset classes will perform tomorrow, let along in 30 years. We use historical data to make guesstimates, but that’s the best we have. By investing in multiple asset classes, we reduce the risk that our portfolio will suffer from the poor performance of a single asset class.
In the final analysis, the correlation among various asset classes, most notably U.S. stocks and Foreign stocks, has increased since 2000. Is this permanent? Doubtful. Correlations go up and down just like returns. The take away for me is that there may be a few asset classes such as natural resources, long-short and global bonds that deserve a closer look than most investors give them. As I’ve mentioned before, here are some good resources for additional reading: