Forecasting energy demand in the United Kingdom
Since the end of the Second World war, United Kingdom energy policy has been highly influenced by beliefs about the adequacy of indigeneous resources, security of supply and macro-economic objectives. Over the same period we observe a growing sophistication in the official projections of energy requirements both in terms of content and technique. An analysis of the methods and models used in the calculation of recent official projections of UK energy demand identifies a number of problems. These mainly concern the size of the price elasticity of energy demand employed in the sectoral models and the modelling of interfuel substitution particularly in the Domestic Sector market for space and water heating. A review of econometric studies of UK demand indicates that the price elasticities of energy demand employed in the Department of Energy models of sectoral energy demand are generally lower than the estimates reported elsewhere. In contrast, official estimates of the own price and cross price elasticities for individual fuels in the Other Industry Sector are broadly consistent with other econometric studies. However the body of work carried out in this area is found to have a number of shortcomings: (i) in general, the problem of autocorrelation has been dealt with inadequately and inappropriately, (ii) it has not been possible to distinguish long run and short run effects satisfactorily, (iii) it has not been possible to distinguish the independent influences of income and price satisfactorily, (iv) models of interfuel substitution have generally performed poorly in the Domestic Sector and (v) the impact of biased disembodied technical change has not been considered in most studies of interfuel substitution. A new study of energy demand in the Industrial and Domestic Sectors of the UK economy attempts to address these problems. The results are mixed. Well established techniques are used to detect and adjust the results for first order autocorrelation. Biases of technical change are estimated and in particular the Domestic Sector interfuel substitution model performs well. However the problem of modelling the dynamic structure of energy demand remains unresolved. In the Industrial Sector, cross section and time series data is pooled in order to distinguish the independent influences of output and price. While the resulting price and output elasticities conform to the values that we would expect a priori, the explanatory power of the model is low. In the Domestic Sector it is possible to develop a model of energy demand with a good predictive performance, however simulation experiments with the model show that the high frequency variation of the dependent variable is mostly explained by the exogeneous temperature variable which itself is difficult to predict. In contrast it is found that relatively simple relationships between overall primary energy demand and key economic and demographic variables can have surprisingly good forecasting properties. This probably arises because of the inertia of systems at this level of aggregation. A severe test is constructed for a number of single equation forecasting models. The parameters of the models are estimated on UK time series data from 1954 to 1973 then the results are used to generate forecasts of the dependent variables over the highly unstable period between 1974 and 1980. Several of the models provide very accurate predictions. The main purpose of the thesis is to assess recent official forecasts of UK energy demand. The single equation model which performs best in the forecasting test is used as a basis for generating projections to the year 2000. The results are compared with the latest official projections. It is found that the official forecasts imply a lower price elasticity than used in the single equation model, a similar relationship between the structural composition of GDP and aggregate energy demand while the implicit income elasticity is higher than our estimate under high fossil fuel price assumptions but lower when fossil fuel prices are assumed to be lower. The simple model yields very similar forecasts to the official projections for 2000 with the exception of one case. With regard to projections of final energy consumption in the year 2000, there are some areas of contention between the bulk of econometric evidence and the implications of the official forecasts but they are small. Most aspects of the fuel mix forecasts are broadly supported by the econometric evidence given the relative fuel price assumptions. It is concluded that the latest official projections represent a fairly realistic appraisal of the likely outcome of an interplay of market forces in the energy market.