Commodities as Portfolio Diversifiers: Non-Correlation Benefits May Be Exaggerated

by: AllAboutAlpha

Every good investment idea has a flaw yet to be identified. With commodity ETFs, it’s contango – future prices that are worth more than today’s spot price. An ETF that rolls over three-month futures contracts risks buying high, with a “negative roll yield.”

With other commodity investments, there’s suspicion that index buyers, primarily institutional investors, simply push up the price of commodities – or at least the futures indexes – with the dominant investor finding himself to be the market, buying and trading against himself.

Nevertheless, commodities have proven a popular alternative asset class, not so much for their potential returns, but their negative correlations against traditional asset classes. According to The Wall Street Journal, inflows into commodities last year were $60 billion. Will those investors be rewarded?

If they are seeking to diversify their traditional stock-and-bond portfolios, probably not, suggest Charoula Daskalakia and George Skiadopoulos, professors at the University of Piraeus. In their paper “Should investors include commodities in their portfolios after all? New evidence,” they write that,

…(commodity) returns are expected to show small or even negative correlation with the returns of assets that belong to traditional asset classes like stocks and bonds. This is because the value of commodities is driven by factors such as weather and geopolitical conditions, supply constraints in the physical production, and event risk that are distinct from those that determine the value of stocks and bonds.

The chart below gives an indication of the risk characteristics of commodities with traditional asset classes.

Commodities seem attractive because, after all, they are real assets, whose values are determined by immediate supply and demand factors. They are not financial assets, whose values depend on some contingent estimate of long-term earnings. With financial assets, market randomness is traditionally managed by mean-variance models (stemming from the work of Nobel Laureate Harry Markowitz) that result in an efficient frontier allocation of stocks and bonds to achieve the highest expected return per unit of risk (or standard deviation).

Can commodities improve that efficient frontier? Daskalakia and Skiadopoulos scout various fences on the frontier:

First, the Markowitz setting may not reflect accurately the gains from investing in commodities since it is founded on two assumptions, i.e., that either the distribution of the asset returns is normal or investor’s preferences are described by a quadratic utility function.

However, asset returns are not normally distributed – Black Swans have yet to be domesticated — and quadratic utility functions break down when investors have achieved wealth sufficient to live a happy life – they’re not willing to take on more risk to obtain more gains. What may seem obvious from observations of everyday life is evidently news to academic practitioners of econometrics, who find that “both these features are not consistent with rational behavior.”

How rational are the econometricians? They “assess the diversification benefits of investing in commodities by eyeballing the position of efficient frontiers.” Not a terribly sophisticated procedure.

Then there’s the commodity universe itself…

The previous literature on asset allocation with commodities has assumed that the investor can invest only in commodity indexes. In practice, this is not the case; instead investors follow different strategies represented by the available menu of futures written on individual commodities.

Still, commodity indexes have a useful, albeit negative, benchmark purpose…

The monthly average return on commodity indexes is lower than stocks and bonds and exhibits higher standard deviation. As a result, the annualized Sharpe ratio is considerably higher for bonds and stocks than commodity indexes. The reported evidence is consistent with previous studies that support that the stand-alone performance of commodity indexes is inferior to other asset classes.

Indexes are one thing. Individual commodities may diverge considerably from their indexed colleagues. “The performance of stocks and bonds is superior to that of all commodity futures but crude oil and gold in terms of risk-adjusted returns,” Daskalakia and Skiadopoulos report.

Back to the 1970s, it seems. But wait. Does adding a “test asset class” – just to try things out — such as commodities improve the risk/reward tradeoff of an already optimized mean-variance portfolio?

MV “spanning” occurs when the MV frontier derived from the augmented investment opportunity set (benchmark assets plus the test ones) coincides with the frontier of the benchmark assets. This implies that the MV investors cannot improve their risk/return trade-off by adding the test assets, regardless of their risk aversion level.

In that test, commodities aren’t worth the effort for traditionally constructed portfolios. But what happens for someone who isn’t fenced in by an efficient frontier portfolio?

The spanning hypothesis is rejected for the two commodity indexes and the majority of individual commodity contracts regardless of the assumed non-MV utility function; the only exceptions occur for futures on cotton (for the assumed exponential utility function) and live cattle.

Bet on commodities if you dare (that is, within your institutionally allocated risk budget). But don’t expect much compensation (and water your cotton and cattle frequently).