Commodity Indices = (Mostly) Energy
In my previous post I showed that commodity indices have become increasingly positively correlated with the U.S. stock market, and today provide little in the way of diversification.
One additional point: despite being billed as diversified measures of commodity returns, commodity indices tend to be driven mostly by energy prices alone.
To illustrate…
Above is the rolling 3-year monthly correlation between the spot price of WTI crude and 4 major commodity indices: Goldman Sachs (blue, ETF GSG), CRB (red), Dow Jones-UBS (green, ETF DJP), and Deutsche Bank (purple, ETF DBC), from 1984.
With the exception of the CRB index, all have exhibited a correlation of +0.85 over the last 3-years (far right of graph). The Goldman Sachs index has been the worst offender with a 3-year correlation to WTI crude of +0.94.
This is mostly a result of the fact that the indices are heavily weighted towards energy: crude, gas, heating oil, natural gas (but also to a smaller degree because many unrelated commodities tend to also move in unison with energy). I realize that I’ve only shown WTI in the graph above, but I think it makes the point just fine.
The only major index to buck the trend is the CRB index which gives a smaller weight to energy. I’m not aware of an ETF that tracks the CRB (readers, please keep me honest here), but the ETF GCC does track the CRB’s close cousin, the CCI.
One last bit of data.
Folks sometimes view commodity indices as a proxy for gold. I think the next graph of the correlation between gold and the 4 major commodity indices dispels that idea (in today’s market) quite nicely.
Having said all of the above, I still use the Goldman Sachs index (ETF GSG) when trading commodities in the MarketSci TAA model.
GS weights its index based on world-production: the average quantity of production world-wide over the last 5 years. Stats be damned, that orderly approach appeals to the inner geek in me.
Happy Trading,
ms
. . . . .
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: Random Stuff | 7 Comments
There is the gsci light if you want less oil.
Hello Sean – good stuff – are there any US traded ETFs or funds that track any of the light indices? A quick google search didn’t turn up anything. michael
Michael,
I have started using. DBA-Agriculture, DBB – Base Metals, DBE – Energy, DBP – Precious Metals. This allows you to make your own tilts or sector rotation using trends (I use 10 month SMA). DBC consists of roughly 20% DBA, 10% DBB, 55% DBE, and 15% DBP. So you can certainly lighten up energy exposure and increase precious metals as an example to get less correlation with equities.-kd
Michael,
I wrote about various issues in selecting a commodity ETF/ETN here: http://tinyurl.com/4xaob3b. I ultimately decided to use a combination of DBA and DBC for the commodities portion of my asset allocation scheme. I also created a small portfolio using an asset allocation scheme at which time I used RJI for the commodity component. Note that much of my thinking about this issue was based on Jim Rogers book, “Hot Commodities”.
Hi ms, great article!
I started to make a predictive model with data from 20 years and many different data sources.
Viewing your article, I find a problem that the correlations they change, I can think of three solutions:
– Use fewer years.
– Find a variable that is responsible for this change
– Maybe this is a correlated relationship but not a causal relationship? Any ideas to detect these cases?
I hope your opnion,
Thank you,
scientists discovered new method to build correlation between 2 variables with no-linear relationship. maybe, it can apply to improve correlations of this post
http://www.sciencemag.org/content/suppl/2011/12/14/334.6062.1518.DC1/Reshef.SOM.pdf