Analysis of the stability of parameters estimates and forecasts in the next-day electricity prices Cover Image

Analiza stabilności ocen parametrów modeli predykcyjnych dla cen energii na rynku dnia następnego
Analysis of the stability of parameters estimates and forecasts in the next-day electricity prices

Author(s): Sławomir Śmiech
Subject(s): Economy
Published by: Wydawnictwo Uniwersytetu Ekonomicznego we Wrocławiu
Keywords: electricity price forecasting; robust modelling; price spike; autoregression models

Summary/Abstract: The purpose of this paper is to investigate the sensitivity of the length of a sample and the impact of outliers on the parameters of predictive models. The analysis was conducted on IRDN, an index of energy prices, which is the general index of the Polish Power Exchange. Taking into account extreme observations, known as electricity price spikes, it was decided to compare the results of tests for models built in the classical way and robust methods. The analysis indicates that taking approximately 600 observations allow to obtain stable parameter estimates and forecasts. Volatility of forecasts obtained by different methods were comparable. At the same time the levels of the values of forecasts in robust and classical methods differed one from another.

  • Issue Year: 2012
  • Issue No: 38
  • Page Range: 135-144
  • Page Count: 10
  • Language: Polish