Normal Backwardation Foiled Going Forward

Includes: CRUD, USO
by: Christopher Holt

According to a forthcoming paper in the Journal of International Money and Finance, there have been significant changes in the risk premia of oil futures since 2005.

Regular readers of AllAboutAlpha will recognize the broader controversy to which these authors are making a contribution. Many writers have contended that the rise of commodity price indexes in the early 21st century has skewed commodity price movements, causing higher prices (or in some variants, greater volatility) both in the futures markets and in the physicals that underlie them. The authors of this new paper, Risk Premia in Crude Oil Futures Prices, suggest that there is some validity to that argument, at least with regard to the commodity with which they’re concerned.

Their own chief contention, though, is that indexing has caused the virtual disappearance of the old-fashioned net-long “normal backwardation” as a source of once-steady average profit for speculators on this commodity.

Crude oil futures began trading on NYMEX in 1983. Volume began a rapid expansion after 2004. The sharpness of the increase is clear from the paper’s Figure 2: Open Interest.

(Click to enlarge)

Theoretical Assumptions

Debate on the underlying theoretical assumptions began to take its present shape more than 80 years ago, when John Maynard Keynes wrote in the second volume of his Treatise on Money (1930) that if producers of a physical commodity want to hedge their price risk, and accordingly sell futures contracts, then they have to be willing to compensate the arbitrageurs who take the other side of that contract, so the futures price ‘should’ be below the expected future spot price, benefiting those net long speculators.

When this theory’s prediction holds true, when the futures price does show such a risk premium in the form of a discounted price, the situation is known as “normal backwardation.” Of course the futures price must converge with the physical spot price on the date of maturity, on basic no-arbitrage grounds. So, if other things are equal, the futures price for a given maturity date will rise over time: delivering on the speculator’s bribe.

Sometimes the theory’s prediction doesn’t hold true, and the situation known as “contango” results. In this situation, the futures price is above the expected spot price, and the necessary convergence suggests that the futures price will fall as maturity approaches. Perhaps the parties who retain title of the physical commodity until maturity are being compensated for the storage costs.

What is the state of play in crude oil futures prices?

The authors of the forthcoming paper, James D. Hamilton of the University of California at San Diego and Jing Cynthia Wu of the Booth School of Business, University of Chicago, say that the index funds as buyers don’t need to be bribed in the way that the speculators Keynes had in mind do or did. The increase in indexed buying is “one explanation for why a long position in futures contracts no longer has a statistically significant positive return.” The indexed buyers are willing to serve as counterparty for commercial hedgers “without the risk compensation that the position earned on average” as recently as the fourteen year period before December 2004.

When Hamilton/Wu plot the 8-week risk premium given their methods, that premium rises dramatically at the start of 2005, but immediately heads south. It has spent most of the time since in negative territory. Their Figure 4 presents this clearly: a great increase of volatility in the premium, along with its effective disappearance as averaged out over the post-2005 period.

(Click to enlarge)

An Affine Model

In more abstract terms, Hamilton/Wu borrow from theories of bond yields, with the theories of “affine models” developed there. An “affine” model [the term itself derives from the geometrical term for transformations that map parallel lines to other parallel lines] is one that treats bond yields and prices for various maturities as arising from a small set of common state variables, adhering to the no-arb principle.

Using the bond theorist’s example, Hamilton and Wu suggest an “affine factor” model of oil futures prices, tied to the mean and variance of the futures portfolio of the arbs.