Why are commodity correlations falling?

From FTAlphaville today:

Banks are selling off their commodity divisions for regulatory reasons but also because commodities are turning out to be less profitable for them than they used to be.

On which note, an interesting development has emerged since banks started winding down their commodity divisions in 2013. According to David Bicchetti and Nicolas Maystre, who wrote a paper in 2012 highlighting increasing correlations between a number of major commodities and indices from 2008 onward, these correlations have now begun to dissipate.

The findings come as a result of an update to their original data set, with the amended charts now looking like this:


FT Alphaville met with the authors last week, who explained that banks exiting commodities could be behind the correlation reversal, which in and of itself is statistically meaningful and therefore worth observing.

Such long-lasting low levels in correlation have not been seen since before the 2008 crisis struck. Back then they were swiftly followed by a spike in correlation, a pattern consistent with what might be expected from a crisis. What is surprising, in the authors’ opinion, is that correlations stayed elevated until at least 2013. The only exception was the breakdown in correlations which occurred in the oil market in the early part of 2011 due to the Libyan crisis.

The two authors argue that the falling correlation is unlikely to be the result of changing fundamentals or US tapering (because it’s still early) and that their research showed commodities are a poor diversification tool given the their propensity to “herd”.

This is chicken and egg stuff. Financialisation played a role, for sure. But so, too, did the 2011 La Nena, MENA crisis, and US QE. For me, though, the primary driver of both the 2008 and 2011 peaks is the correlation with China. Both periods were the very peaks of Chinese demand growth.

It’s classic model for a bubble. A kernel of truth leading to an overblown and ultimately destructive market response culminating in hysteria and crazy correlations.

As the great “super cycle” bubble unwinds on the reset in Chinese growth and the overblown supply response you can, ironically, expect correlations to fall much further.

Houses and Holes


  1. migtronixMEMBER

    Thanks for this piece HnH, all sorts of correlation I’ve seen break- just last night AUD was moving in the opposite direction to the other commdolls.

    Speaking of commodities wheat, soybean and cocoa have a great run this year

  2. I know its ancient history, but I think its worthy to remember that mineral commodity Houses, (e.g. BHP, Rio, North, Western Mining, etc) had a two decade negative correlation under CAPM; losing approximately 10% pa. In this context, to contrast the average 70-90% positive correlation between commodities and other asset classes over this past decade. A relationship reverting to the mean I suspect.