We propose a discussion index model (Stock and Watson, 2002) to fore- cast electricity demand for one hour to one week ahead. The model is particularly useful as it captures complicated seasonal patterns in the data. The forecast performance of the proposed method is illustrated with a simulated real-time experiment for data from the Pennsylvania- New Jersey-Maryland Interchange.

diffusion index forecast, electricity load, seasonality
Hypothesis Testing (jel C12), Time-Series Models; Dynamic Quantile Regressions (jel C22), Time-Series Models; Dynamic Quantile Regressions (jel C32)
hdl.handle.net/1765/8001
Econometric Institute Research Papers
Report / Econometric Institute, Erasmus University Rotterdam
Erasmus School of Economics

Rotger, G.P, & Franses, Ph.H.B.F. (2006). Forecasting high-frequency electricity demand with a diffusion index model. (No. EI 2006-38). Report / Econometric Institute, Erasmus University Rotterdam. Retrieved from http://hdl.handle.net/1765/8001