Modelling Sector-wise Electricity Demand in Sri Lanka: An Application of Vector Autoregressive Model
K. A. N. K. Karunarathna *
Department of Mathematics, Faculty of Science, Eastern University, Chenkalady, Sri Lanka.
K. Vithyasangaran
Department of Mathematics, Faculty of Science, Eastern University, Chenkalady, Sri Lanka.
*Author to whom correspondence should be addressed.
Abstract
The knowledge about the current and future electricity demands is necessary and much useful in planning and decision making for giving proper electricity supply. This study aimed to identify the pattern and to model the sector-wise electricity demand. Sector-wise (domestic, industrial and commercial) electricity demands from 1970 to 2015, were collected from central bank annual reports. The series of second order differences of demand in each sector, which showed stationarity, was modelled by using Vector Autoregressive (VAR) models.
VAR model of order 3 (VAR(3)) could explain the behaviour of series of second order differences. This model showed 85% of minimum accuracy in prediction. There could be seen an increasing trend of demand in all the sectors. Up to late part of 80’s, a linear trend could be observed in each sector. Demands in domestic and commercial sectors were almost the same up to around 1988. After year 1988, the demand in domestic sector was higher than the demand in commercial sector. Up to about 2000, demand from industrial sector was higher than the other two sectors. During the period from 2000 to 2006, demand of domestic and industrial were almost the same. But after 2006, demand in domestic sector has gone up than the demand in industrial sector.
Keywords: Electricity, forecasting, VAR, demand.