Simulation of the Evt Model with Copula Functions

The literature has earlier examples that pointed out the weaknesses in traditional Monte Carlo simulations in modelling electricity prices. For example, Lucia and Schwartz (2002) applied EVT to electricity prices to assess the level of tail fatness in series followed by comparison of the Monte Carlo simulation results based on Gaussian and student-t distributions. Their simulation results improved upon the ones from the Gaussian distribution, as the Student-t price patterns resemble more closely the underlying price pattern of electricity prices. This study has showed that the normality assumption that researchers and practitioners often make in their simulation or valuation method is not appropriate and prone to lead to erroneous conclusions. This further strengthens the view taken in this section in simulating the electricity prices data modelled with the assistance of EVT. In this section, the residuals of the GPD model are simulated with copula functions. As previously mentioned, copula functions allow modelling of price and volatility dependencies, which exist in the regions of NEM. The copula functions also provide a multivariate approach to modelling electricity prices in NEM.