Information Flows in the term structure of commodity prices

Relying on conditional entropy and on the notion of information transfer, we investigate price relationships in the most important commodity futures market: the American crude oil market. We first show that the information shared by futures contracts with different delivery dates increases during the period under scrutiny (i.e. 2000-2011). This is especially true for intermediate maturities. When focusing on information transfer, on average on the whole period, it appears that short-term maturities emit more information than long-term ones. This is consistent with the normal functioning of a futures market. A dynamic analysis however reveals that the relative importance of information flows emerging in the far end of the curve (for long-term maturities) arises as integration progresses in the crude oil market. The transmission of shocks from the paper to the physical markets is thus facilitated. Last but not least, the direction of prices moves becomes less stable as time goes on. On the theoretical point of view, these findings raise questions about the segmentation theory and the Samuelson effect.