| Summary: | 碩士 === 東吳大學 === 經濟學系 === 97 === This study discusses the forecasting models of electricity demand in Taiwan, which are ARIMA model and Regression model. From this study, we find that lagged periods are inconsecutive in ARIMA model. For Regression model, it is necessary to cope with variables of different integrated order and autocorrelation of disturbance.
We reserve some data as out of sample to test the forecasting accuracy. The result shows that the Regression model is slightly better than the ARIMA model, but the difference is small. By Theil’s inequality coefficient, the forecasting ability of these two models are equal.
Besides, we calculate the income elasticity and the price elasticity of electricity. Both are smaller than 1, so electricity is a necessary good, and its price is inelastic.
|
|---|
