Should LNG Pricing Be Oil-Indexed or Gas-Indexed? A Perspective from the Predictability of Value at Risk (VaR)

One of the main reasons for retaining the oil index price is that the crude oil price series is less volatile than the natural gas price series. In this study, we claim that the predictability of volatility is more important than the size of volatility. Thus, three simple and commonly used Value at Risk (VaR) models-the Historical simulation (HS), Mean-Variance (MV), and GARCH models — are applied to calculate the 95% VaR for gas and oil price indices. Further, the back-testing is used to access the performance of VaR method. We find that the violation duration of HH index is independent based on GARCH model, and only slightly auto-correlate based on HS or MV method. But the violation duration of Oil indices, are significantly auto-correlated. Furthermore, we apply the ACD model to fit the series of violation duration that are auto-correlated. An interesting finding is that the ACD coefficient estimated from oil's violation duration is almost larger than it from HH's violation duration. The larger ACD coefficient reveals that while the violation event occurs earlier than did the expected value (violation duration deviated from the long-run equilibrium), it will takes longer time to convergence to the expected duration. The violation duration can be forecasted by using the ACD model, however, to avoid the occurrence of VaR violation will cost more for oil index than for gas index