Predicting dependent electricity price spikes through copula functions
MetadataShow full item record
The aim is to forecast the occurrence of extreme prices (the so-called spikes) in the Australian electricity markets from half-hourly electricity spot prices. Specifically, we are interested in the simultaneous occurrences of such spikes in two or more interconnected markets. In order to do so, we propose a copula-based econometric framework for the prediction of co-spike probabilities in interconnected markets. The model allows for a flexible choice of the marginal link functions, while the dependence structure among marginal occurrences is described by means of a copula function. Compared with benchmarks assuming independence among markets or standard multivariate choice models, our approach reveals itself to be more conservative on predicting simultaneous occurrences of extreme prices.