All posts from in,

Can Investors Achieve Commodity Exposure Via Equities?

Can Investors Achieve Commodity Exposure Via Equities? by Wesley R. Gray, Ph.D., Alpha Architect, Author of Quantitative Value: A Practitioner’s Guide to Automating Intelligent Investment and Eliminating Behavioral Errors.

This past year we examined the possibility of replicating commodity exposure via equities. The project was spurred by an insightful research report from MSCI, which showed some impressive results. Other research outfits have proposed similar concepts. The figure below, taken from the MSCI report, highlights how well the MSCI Select Commodity Producers Index replicates various commodity indices over the 2010-2012:

We replicate these results and come to similar conclusions:Correlations for commodities and commodity-related-equities are >.8 from 2010-2012.

However…and this is a big however…

When we look at a longer out-of-sample period, from 1991 to 2014, correlations are much lower (the best versions of our algorithms can get the correlation in the .6-.7 range after intense data-mining). The executive summary below is from a 125-page internal report we did on commodity via equity replication. The correlation figures represent the full-sample correlations between the underlying commodities and some of our top replication techniques. Clearly, the evidence below suggests that we should be skeptical of claims that commodity exposures can be effectively replicated via commodity-related-equities. Especially, when the sample period analyzed is short.

Understanding Commodity Replication via Equities

Accessing commodity exposures can be complicated.

Consider oil exposure: Buy oil futures? The oil ETF? Oil stocks? Each of these option has pros and cons.


Futures appear straight forward, and require less margin than equities; however, these contracts trade in large notional amounts (eliminating the option for retail investors), incur transaction costs (potential roll costs and other transaction costs), and trading futures should not be viewed as a buy-and-hold investment (e.g., see research here and a>