Explaining the so-called "price premium" in oil markets

This paper explores the information content of several variables on the so-called "oil price premium over fundamentals". We define this premium as the difference between the market oil price and the estimated price consistent with the OECD's relative industry stock level. By using Granger causality tests and extended regressions we test the systematic ability of a broad set of variables to explain the premium. We find that speculation in the oil market - measured by non-commercial long positions - can improve the traditional model, reducing the premium significantly during some parts of the sample.