Detecting abnormalities in the Brent crude oil commodities and derivatives pricing complex

Recent rapidly rising and volatile energy commodities prices and financial price manipulation scandals have brought the pricing mechanisms of crude oil derivatives to the fore of both popular press and policy initiatives. Among the most important of such commodities is Brent Crude. Brent Crude and its complex of derivative products make Brent Crude potentially more opaque and thus susceptible to price manipulation than other commodities. In spite of the importance of Brent to the world economy and world energy prices, and its complex of derivative pricing, relatively little work has been done to explore the potential for, and evidence of, price manipulation in the Brent Crude complex. This paper seeks to address this lack by proposing a method to test whether price squeezes have occurred in Brent Crude. This paper builds on previous work which proposed an a priori test for evidence of manipulation and the theory of storage. Previous work (Barrera-Rey and Seymour 1996) posited that the very close-to-delivery end of the forward curve for Brent should not be simultaneously in contango and backwardation, while other work (Geman and Smith 2012) proposed using an econometric prediction and a model based on the theory of storage to detect manipulation in commodity markets. Our work builds on these approaches by developing a more detailed model of calendar spreads in the Brent Crude complex. In Brent, a particular area of potential manipulation is from the relatively illiquid and more opaque physical OTC forward market (where prices are ‘assessed’ by Platts during a short ‘window’ of time) and the more liquid ICE futures market. Our model relates prompt ICE futures calendar spreads to prompt-over-dated OTC forward spreads. The model then tests whether the a priori indicators of manipulation as suggested by Barrera-Rey and Seymour are statistically consistent with the process which drives spreads historically. We find that in most all cases, the indicated period of manipulation is statistically different. We further ...