Evaluating robust determinants of the WTI/Brent oil price differential: a dynamic model averaging analysis

We investigate the robust determinants of the WTI/Brent oil price differential by employing a time-varying framework. To achieve this, a Dynamic Model Averaging framework is used, considering monthly data over the period 1994:1-2021:3. Our results suggest that the convenience yield, the global economic activity index and the government bond yields act as the main factors that exercise a persistent and significant impact, for the largest part of the study period, although at different magnitude. More importantly, though, we show that at different time periods there are additional factors that exercise a significant impact on the oil price differential, such as refining constraints, stock market volatility, trading volume and geopolitical risk. Thus, unless a dynamic modelling framework is employed, the full spectrum of the related effects cannot be revealed. A series of tests confirm the robustness of our findings. Several policy implications of these results are also discussed.