Probability Distribution of Electricity Demand Forecast for the City of Karachi

A prerequisite for developing the planning and demand management policies for energy is the appropriate projections of future demand. Capacity targets may be set particularly more efficiently in the light of a knowledge of the probability of shortages. However, such probabilities are usually not available. Such probability distributions are generated in this study. More specifically, a simple four-equation model has been developed to make projections of average and peak demand for the city of Karachi in the year 2000. Three of the four equations have been estimated using Full Generalised Least Squares with Prais-Winsten transformation in order to correct serially correlated errors. The estimated model, after making appropriate tests for hetroscedasticity, has been put to recursive bootstrapping to generate probability distributions of average and peak demand in order to assess the extent of uncertainty in point projections. Bootstrapping has been used because of the limitations of the conventional method with regard to imposing a pre-specified stochastic structure on the error term and assuming the knowledge of the values of the exogenous variables for the forecast period with certainty. Probability distributions are based on 1000 random samples. The results indicate that under quite plausible assumptions, the extent of uncertainty remains significant which should be taken into account for future policy planning.