Investigation on the impact of news on volatility in load time series

Abstract

Load forecasting is of great importance in power system. With the development of the electricity market, the precision of load forecasting is paid more attention to than ever before. The investigation on characteristics of load time series is beneficial. In this paper, by employing the ARCH type models, the asymmetric characteristics of load series are discussed and an investigation on the impact of news on volatility is represented. Several interesting results are documented. The utility of the ARCH-type models with different empirical conditional distributions in describing time-varying volatility are demonstrated and the reverse leverage effect is reaffirmed by the asymmetric parameters estimation. Owing to the news impact curve (NIC) analysis, the responses in volatility of time series to the sign and the magnitude of the idiosyncratic shocks are highlighted. All the results are mutually confirmed in spite of different model specifications. The empirical results also show the satisfying forecasting ability of the models in view of some standard summery statistics. The asymmetric GARCH models may be a valid and promising method for load time series analysis.