Forecasting diesel consumption in Sri Lanka using multivariate time series models

Abstract

In the modern trends of industrialization and development, energy has become one of the most important aspects of every economy in the world. Among the energy sources that are available today, diesel has a considerable consumption for various activities such as production of goods and transportation in many countries including Sri Lanka. Using monthly data of ten explanatory variables (January 1998 to July 2018) vector error correction model of order 2: VECM (2) was developed to model monthly consumption of diesel in Sri Lanka. The diesel consumption has been increasing due to various activities. The most significant influential variables are Exchange rates of USD to LKR, Merchandize Imports, Number of Tourists Arrivals, National Consumer Price Index, and Electricity Power Generated. The errors of the model were found to be white noise. The percentage errors of the fitted data using the VECM (2) model for both trained and validated set vary from -8.4 to +8.5%. Further it was found that Exchange rates of USD to LKR, Merchandize Imports, and Number of Tourists Arrivals show significant long run positive association with diesel consumption while National Consumer Price Index and Electricity Power Generated indicate significant long run negative relationship with diesel consumption in Sri Lanka. This model is suitable only short term prediction and it is recommended to develop the model so that it can be used for long run prediction. Nevertheless, the model provides the analyst with the ability to make decisions using various predicted intervals with different membership values by controlling the explanatory variables.