We study the state-dependent trading behavior of financial institutions in the oil futures market, using structural vector autoregressions with Markov switching in heteroskedasticity. We consider two states of the world: tranquil and turbulent. We decompose the observable time-varying price volatility during the period 2006M6–2016M5 into changes in the slopes of traders’ demand curves and into changes in the variability of their demand shocks. We find that the downward-sloping demand curve of intermediaries steepens significantly during crises times and that the variance of their demand shocks doubles. These findings suggest that the futures pricing of financial institutions is highly nonlinear and raises the hedging costs of producers and consumers of oil when volatility is high.
Keywords: Asset pricing, Trading behavior, Financial institutions, Vector autoregressions, Market microstructure, Oil markets