Historical Correlation Between the S&P 500 and Major Indices

Hickey and Walters (Bespoke) submit: We recently decided to see how correlated the S&P 500 is with various asset classes and if they were becoming more or less correlated in recent years. We first calculated the weekly percent change going back to 1988 of the S&P 500, FTSE 100, Nikkei 225, MSCI World, MSCI Emerging Market, 10-Year Treasury Note, Oil and Gold. We then calculated a 1-year rolling correlation between the weekly percent changes of the S&P 500 and each market. The historical correlation charts are shown below.

While equity indices globally have become more correlated with the U.S. in recent years (see MSCI World, Emerging Markets, FTSE 100 and Nikkei 225 charts), the S&P still has a relatively low level of correlation with non-equity related assets.

click to enlarge

Comments

[inbox:]

Thank you for investigating these correlations. This is an important topic.

I think you are oversimplifying some cases by not controlling for common variables. Consider the case of the S&P 500 vs 10-year Treasury bonds. When I calculate the simple correlation of those prices over 10 years, I get -19%. But when I control for short-term interest rates, I get +40%.

Likewise, the correlation over the most-recent 2 years is -30%. But controlling for short-term rates, it becomes +57%.

Not only is are the controlled correlations different, they actually have opposite signs from the uncontrolled correlation! That is significant. The controlled calculation reveals a relationship we might otherwise miss.

F.Y.I., I used a continuous bond futures price as a surrogate for the 10-year bond price. And I used LIBOR as my short-term interest rate. Also, the partial correlation calculation is easy to implement. You can read about it on Wikipedia.

2007 Jun 11 09:36 PM