Letter To The Editor: Creeping Correlations

March 15, 2007

Commodities still work when it matters most: when the markets stumble.


[Editor's Note: Nik Bienkowski is head of listings and research for ETF Securities, which markets commodity exchange-traded funds (ETFs), primarily in Europe.

Dear editor:

I read with interest the article posted last week on IndexUniverse.com (Creeping Correlations, by David Krein). It sparked a pair of important questions for me:

(i) Are correlations really rising with previously uncorrelated assets; and

(ii) If so, does this increased correlation hold during market downturns, when an investor relies on the low correlation of other asset classes?

My queries of the article's conclusions were justified: The results of my sample do not support the view that the correlation between equities and commodities is rising. In fact, sampling various periods showed that commodities not only have maintained their low correlation, but they also maintained the lowest correlation among the indexes analysed. Further, when equity markets performed poorly, commodities performed the best relative to the other asset classes, while the correlation between equities and commodities remained the lowest.

Correlations With Commodities Still Remained Low

Factors that affect short-term versus long-term returns

Whilst it is possible that correlations may rise over the short-term - for example, commodity market and equity market returns have both been strong over the last 18 months or so -- the correlation should not rise over the longer term if the asset classes really are driven by different factors.  It is not hard to make a case why commodities should be uncorrelated to equities, as commodities are affected by many different factors including exploration success, technological advances, supply-demand (and the time lags to respond to supply/demand changes), the weather, and the depletion of finite resources. 

Short-term or long-term correlations?

Index investors, specifically passive investors, when forming portfolios of uncorrelated assets, are generally making a call about the long-term fundamentals of an asset class relative to others.  Therefore, whilst it is important to understand the dynamics of non-static correlations, it is important to consider past and future expectations for correlations over a time period that is consistent with the investor's goals.  Unless the investor is active and trading frequently (where even positive short-term correlations may be viewed as good), then low-to-negative correlations need to be viewed with a medium- to long-term horizon.

Commodities have maintained their low correlation

When looking at variations in correlation, the best way to view potential shifts in correlation is by looking at rolling correlation charts, rather than examining tabular results that could be affected by the time period chosen.

The charts below examine the 2- and 5-year rolling correlations of various indexes with the S&P 500, including foreign stocks, hedge funds and commodities. They show that:

·       correlations are more volatile over shorter periods;

·       in each case, the correlation between commodities and equities was the lowest of any group studied;

·       the correlation between commodities and equities has consistently remained low, between 0.0 and  0.2; and

·       the correlations between equity indexes has risen over time.

Note that these results do not show that the correlation of asset classes with lowest correlation increased the most over the past five years.


Find your next ETF

Reset All