Commodities, equities and portfolio allocation

The BIS has published a new paper titled "On the correlation between commodity and equity returns: implications for portfolio allocation" which can be downloaded here.  This is quite different from the usual fare from the BIS, which has centred on macro-financial issues in the last few years.  Perhaps the Authors Lombardi and Ravazzolo are looking for a new career in financial markets.

While the paper is very wonky in terms of econometrics used, there are some useful takeouts without going into the equation-heavy detail and is an interesting read for those immersed in equity markets and portfolio management.

First the authors show that the correlation between equities (in this case the S&P500) and commodities (S&P GSCI index) has been much more positive and significant since September 2008.  This is shown by not just using rolling correlations of different durations, but by using something called Dynamic Conditional Conditional Correlation model (DCC), which allows for time-varying correlations.

This model is then used to provide joint forecasts of commodity and equity returns.  The authors provide both point forecasts and density forecasts, which have additional information about probabilities of outcomes.  The results suggest that equities provide predicative information of commodities, but not the other way around.

In the final part of the paper, the authors specify a portfolio where an investor can invest in either equities, commodities or risk free bonds (taken as the Fed Funds rate).  The authors suggest that active portfolio allocation based on model results to equities/commodities can outperform passive strategies, or those models which assume returns are a random walk. Calculation of Sharpe ratios also show that contrary to some opinions, commodities do not act as a hedge in portfolios.  If anything, commodities have been increasingly pro-cyclical in portfolios.

To be clear, when this paper talks about commodities, its mostly talking about energy given the S&P GSCI index is overwhelmingly oil and fuel.  Perhaps an interesting extension to this paper would be to look at metals or agricultural commodities separately to see how they differ from the GSCI.

I think this approach is a nice supplement to more detailed fundamental analysis on commodities and equities, rather than an alternative approach.  To be sure, everything that drives the changes in correlation between equities and commodities is in fact exogenous to this framework.  In some ways that is good, as it establishes relationships that aren't dependant on complicated economic variables. But this pure econometric approach is not a substitute to understanding broader macro trends.