Abstract/Summary

In our analysis of electricity price series from Australia’s National Electricity Market (NEM), we employ the indicator saturation (IS) approach to simultaneously model the stylised facts of electricity prices, including extreme spikes, seasonality, level-shifts, and autocorrelation. The standard modelling methods described in the literature tend to use regime-switching models to cope with these characteristics, but these models cannot fully reflect the stylised facts of interest. Using a range of model-evaluation tools, our analysis finds that the IS method outperforms the regime-switching models in various settings. In addition to the statistical superiority of this approach, we detail the relevance of our findings to policymaking in the NEM and provide recommendations for the development of the electricity markets in Australia.