Determining New Oil Market Predictors Under Model Uncertainty

It has recently been suggested that speculation is now playing an important role in daily price movements of global oil prices. This raises the question: what are important drivers of price changes given this new 'speculative' regime? We identify new factors of the oil market related to speculation by fitting Subset Vector Autoregression models with Exogenous variable (SVARX) and rank them by importance. Further, to account for model uncertainty and to obtain robust parameter estimation in this exploratory study, we apply a bootstrap model selection procedure.