This paper investigates the behaviour of spot prices in eight energy markets that trade futures contracts on NYMEX. We consider two types of models, a mean-reverting model, and a spike model with mean reversion that incorporates two different speeds of mean reversion; one for the fast mean-reverting behaviour of prices after a jump occurs, and another for the slower mean reversion rate of the diffusive part of the model. We also extend these models to incorporate time-varying volatility in their specification, modelled as a GARCH and an EGARCH process. We compare the relative goodness of fit of the different modelling variations both in sample, using Monte Carlo simulations, as well as out-of-sample, in a Value-at-Risk (VaR) setting.
Our results indicate the presence of a "leverage effect" for WTI, Heating Oil and Heating Oil-WTI crack spread, whereas for the remaining energy markets we find the presence of an "inverse leverage" effect. Also, the addition of the EGARCH specification for the volatility process improves both the in-sample fit as well as the out-of-sample VaR performance for most energy markets that we examine.