A two-factor model for electricity prices with dynamic volatility

The wavelet transform is used to identify a biannual and an annual seasonality in the Phelix Day Peak and to separate the long-term trend from its short-term motion. The short-term/long-term model for commodity prices of Schwartz & Smith (2000) is applied but generalised to account for weekly periodicities and time-varying volatility. Eventually we find a bivariate SARMA-CCC-GARCH model to fit best. Moreover it surpasses the goodness of fit of an univariate GARCH model, which shows that the additional effort of dealing with a two-factor model is worthwile.