Modelling technical progress: an application of the stochastic trend model to UK energy demand

The precise role of technical progress in estimated energy demand functions has not been well researched. Traditionally a deterministic time trend has been used, implicitly assuming technical progress continues as a fixed rate over time. In this paper, the structural time series model is employed allowing for a stochastic time trend and stochastic seasonal dummies. Therefore, technical progress and seasonal variation are treated as unobservable components that evolve over time. The conventional deterministic trend model is a restricted case of the structural time series model and found not to be accepted by the data for a number of energy types. Energy demand functions for a variety of energy types are estimated for the UK using unadjusted quarterly data. It is found that technical progress in energy usage does not always exhibit a deterministic trend pattern as the conventional model assumes. It often fluctuates over time and is likely to be affected by a range of exogenous factors but also by changes in energy prices (and possibly income also).