Since oil is a non-renewable resource with a high environmental impact, and its most common use is to produce combustibles for electricity, reliable methods for modelling electricity consumption can contribute to a more rational employment of this hydrocarbon fuel. In this paper we apply the Principal Components (PC) method to modelling the load curves of Italy, France and Greece on hourly data of aggregate electricity consumption. The empirical results obtained with the PC approach are compared with those produced by the Fourier and constrained smoothing spline estimators. The PC method represents a much simpler and attractive alternative to modelling electricity consumption since it is extremely easy to compute, it significantly reduces the number of variables to be considered, and generally increases the accuracy of electricity consumption forecasts. As an additional advantage, the PC method is able to accommodate relevant exogenous variables such as daily temperature and environmental factors, and it is extremely versatile in computing out-of-sample forecasts.
Electricity Load curves Principal components Fourier estimator Constrained smoothing estimator Temperature Non-renewable resources Hydrocarbon fuels Environment