The repricing of risk in the capital and commodity markets may be intimidating, but it's a natural, recurring process, and one that often brings fresh opportunity (along with new and so perhaps unexpected risks) for strategic-minded investors.
Periodic rebalancing as a general rule is a good idea, perhaps more so than usual these days given the rise in volatility in some asset classes and the growing sense that more than the usual demons haunting the markets and the economy may be lurking in the shadows. All of which provides a timely excuse to update the correlation trends between stocks, bonds, REITs and commodities if only to see how the recent turmoil in markets has reshuffled the relationships between the asset classes.
To keep things manageable, we've crunched the correlations from a U.S. stock market perspective (see chart below by clicking for larger image). The decision doesn't mean that looking at correlations from the vantage of bonds or REITs or commodities is unproductive. Indeed, a full and prudent study of correlations demands considering all the angles. But in the interest of brevity, today we look exclusively at how correlations have evolved vis a vis U.S. equities, as represented by the Russell 3000, which is a broad measure of the market.
Before we start analyzing the trends, let's first define some terms. The chart above profiles 36-month rolling correlations based on monthly total returns for the respective markets. For example, the correlation between U.S. stocks and commodities for January 2008 comes from a correlation derived on the previous 36-month total returns for each asset class.
Correlation measures the relationship between two data series, in this case monthly total returns. A correlation of 1.0 equates with perfect positive correlation, meaning that the two markets are effectively one and the same, or at least highly similar. A correlation reading of 0.0 is no correlation, and a correlation of -1.0 is perfect negative correlation. And, of course, there's a strong case for building portfolios by mixing asset classes with low and negative correlation. The devil's in the details, but as a general rule this one carries a lot of weight in our book.
Okay, so what is the above chart telling us? The first lesson is that correlations evolve, which means that the diversification opportunity of any given asset class relative to other varies over time. Consider, for instance, that correlations between U.S. stocks and REITs (red line) have been rising since 2002. As of last month, the trailing 3-year correlation between the two was 0.62, up sharply from 0.19 five years ago.
Another example: the correlation between U.S. stocks and foreign government bonds in developed markets (based on unhedged dollar returns) has fallen dramatically and relatively quickly (pink line). As of last month, the two posted a negative correlation: -0.35 vs. a modestly positive 0.31 in November 2006.
If we step back, we can see four broad trends in the data:
The first trend: Correlations between U.S. stocks with foreign stocks in both developed and emerging markets have been slipping, albeit from the high levels in the recent past, which was a byproduct of the powerful bull market in equities around the world that's now just a memory.
The second trend: Rising correlations between U.S. stocks and REITs; and U.S. stocks and high-yield bonds.
The third trend: Falling correlations between U.S. stocks and U.S. bonds; and U.S. stocks and foreign bonds.
The fourth trend: The low but still modestly positive correlation between U.S. stocks and commodities seems to be oscillating within a range so far in the 21st century. A fear that the securitization of commodities via ETFs, ETNs and mutual funds in recent years would destroy commodities' diversification value isn't supported by the correlation data, at least so far.
The past, of course, is always clear - it's the future that's always the mystery. In fact, the only thing that really matters in portfolio design is the future path of correlations. Alas, can only guess. That leaves us with the quantitative crumbs, namely: Does the past provide any insight into the future on the matter of correlations? Some, although caveat emptor still applies. Indeed, we know that correlations are constantly changing, though not necessarily in smooth and predictable trends.
We'll close with one speculative reading of the trend: the diversification benefits of owning foreign and domestic stocks will modestly improve in the coming months and years, which is another way of expecting the correlations between the two will continue falling. Again, that's just a guess, but knowing that the correlations in this case have been quite high in the recent past, and that correlations evolve, there's a case for arguing that lower correlations are coming. Maybe.