Robust smooth transition threshold autoregressive models for electricity prices

Abstract

In this paper we suggest the use of robust STAR (Smooth Transition AutoRegressive) processes to model and forecast electricity prices observed on deregulated markets. The robustness of the model is achieved by extending to time series the M-type estimator based on the polynomial weighting function first introduced for independent multivariate data. The robust M-STAR estimator can be considered as a generalization of the robust SETAR estimator [1], because in STAR processes the change from one regime to another is ruled by a smooth function rather than by a fixed threshold. The main advantage of estimating robust STAR models is the possibility to capture two very well-known stylized facts of electricity prices: nonlinearity produced by changes of regimes and the presence of sudden spikes due to inelasticity of demand. The forecasting performance of the model is assessed through an application to the Italian electricity market (IPEX). By means of prediction performance indexes and tests, robust and non-robust STAR models for electricity prices are compared.