Drivers of real oil prices have been explored extensively in the literature with little consensus. Using a new identification scheme based on forecast error variance, we identify oil-specific demand, demand, and oil supply shocks that maximize the sum of forecast error variance of three variables explained by their respective shocks. The estimation, with the sample period until 2007, suggests that the three identified shocks have similar effects as in the early literature, with oil-specific demand shocks playing a prominent role. However, in the post-crisis period supply shocks have emerged as a source of short-run increases in oil prices, and demand shocks do not have a long-run effect on prices, unlike in the pre-crisis period. By further including risk in the model, we show that the importance of supply shock in driving oil prices in the short run is not driven by global risk. These estimates overwhelmingly suggest zero short-run supply elasticity - a matter of debate in the recent literature. Aside from oil-specific demand shocks, six episodes (including COVID-19) and a time-varying VAR with stochastic volatility identified based on forecast error variance suggest that other shocks, in particular supply shocks, have also played a significant role in driving oil prices in different episodes which cannot be ignored while evaluating the oil price dynamics.