Correlation and Volatility: Some Healthy Signs

Includes: SPY, TBT, TLT
by: Alan Schram

There is a strange divergence between correlation and volatility recently. The two traditionally move together: the more volatile markets are, the more correlated they tend to be. During the financial crisis, with global uncertainty spreading worldwide, volatility spiked. Correlation across asset classes soared along with it to historically high levels.

That made sense. Since then, however, volatility has declined as fears of systemic collapse eased, but correlations have not come down. Right now, correlation among asset classes remains very high, and that pertains even to disparate assets such as commodities and REITS, while volatility is declining.

During the last 15 years, correlation coefficients have been reasonably low. Take emerging markets as an example: China has been the least correlated to the S&P 500 with a coefficient of 0.41, the Nikkei 225 coming in next at 0.57 and the BRIC index coming in third at 0.66 (a correlation coefficient of 1 would imply that the S&P 500 and the foreign index move in lockstep, 0 means that there is no relationship between the two, and a negative reading means they move in opposite directions). Now, the average correlation among the 45 markets in the MSCI World Index is above 0.7, up from about 0.3 a decade ago.

So global markets are moving in unison. That is highly unusual, and it also has major consequences. Correlation between asset classes is central to asset allocation decisions. People use assets with low correlation to reduce the volatility of their portfolio, believing (wrongly) that it also reduces their risk.

Even sophisticated investors use assets with negative correlation as portfolio hedge. Stronger correlations frustrate those efforts and make it more difficult for such investors to distinguish themselves. Indeed, the standard deviation (or variation in returns) for large cap U.S, mutual funds declined to 4.1% in the second quarter of 2010 (per data by Lipper and Bloomberg). That was the lowest in a decade.

There are several likely reasons for this phenomenon. First, small investors are abandoning the equity markets, leaving behind ETFs and large quant investors with high-frequency trading platforms. Those tend to operate on similarly designed formulas, and thus move together.

Second, as the world becomes more globalized, large corporations are now bringing in more international revenue, making them more interconnected and more dependent on the global economy. Now, a downturn anywhere in the world impacts almost all large companies, regardless of their domicile. For example, the credit crunch was a global event, and many investors found their geographic diversification of little use.

However, while correlations of US stocks and other asset classes are high, the correlation with US Treasuries has recently turned down. This is most likely due to the decline in fear levels. Investors tend to flee to treasuries as a safe haven, so historically correlations between stocks and bonds shift according to people’s fears and hopes.

For example, during the bull market from 2002 to 2007, when the S&P 500 nearly doubled, the correlation between the S&P 500 and 10-year Treasuries averaged 0.15. But the correlation jumped to 0.79 in the summer of 2007, as U.S. sub-prime mortgage losses roiled credit markets. It rose further to 0.83 in the Fall of 2008, after the Lehman bankruptcy.

Recently, U.S. shares are moving more independently of the bond market, with correlation levels substantially lower than the peaks established in the Great Recession. This is an indication that profits and valuations are again guiding investors more than concern about the economy, particularly now that more companies are posting earnings growth.

Of course correlation coefficients are not indicators of value. But at least the weaker connection among stocks and bonds is one reassuring sign.

Disclosure: None