What drives oil prices? - A Markov switching VAR approach

This paper constructs a five-variable Markov switching vector autoregressions (Markov switching VAR) model based on oil prices, oil aggregate supply, oil aggregate demand, global oil inventory, and oil speculative demand. Specifically, we build this model to study the impact of different oil shocks on oil prices from May 2000 to April 2020 and analyze the driving factors of oil prices under different regime conditions. Empirical results show that the oil inventory and speculative demand have more significant effects on the oil price fluctuation than the oil aggregate supply and demand. Even though the regime probability indicates that the oil market is relatively stable, some unexpected non-economic factors may become the fuse to disturb market order. Furthermore, we find that a single factor can not drive the oil price fluctuation. Multiple factors from the cumulative effects on the oil price fluctuation and the intensity of these factors change under different regimes.