Accounting for asymmetric price responses and underlying energy demand trends in OECD industrial energy demand

This paper explores the way technical progress and improvements in energy efficiency are captured when modelling OECD industrial energy demand. The industrial sectors of the developed world involve a number of different practices and processes utilising a range of different technologies. Consequently, given the derived demand nature of energy, it is vital that when modelling industrial energy demand the impact of technical progress is appropriately captured. However, the energy economics literature does not give a clear guide on how this can be achieved; one strand suggests that technical progress is ‘endogenous’ via asymmetric price responses whereas another strand suggests that it is ‘exogenous’. More recently, it has been suggested that potentially there is a role for both ‘endogenous’ and ‘exogenous’ technical progress and consequently the general model should be specified accordingly. This paper therefore attempts to model OECD industrial energy demand using annual time series data over the period 1962 -2010 for 15 OECD countries. Using the Structural Time Series Model framework, the general specifications allow for both asymmetric price responses (for technical progress to impact endogenously) and an underlying energy demand trend (for technical progress and other factors to impact exogenously, but in a non-linear way). The results show that almost all of the preferred models for OECD industrial energy demand incorporate both a stochastic underlying energy demand trend and asymmetric price responses. This gives estimated long-run income elasticities in the range of 0.34 to 0.96; long-run price-maximum elasticity in the range of -0.06 to -1.22; long-run price-recovery elasticity in the range of 0.00 to -0.71; and long-run price-cut elasticity in the range of 0.00 to -0.13. Furthermore, the analysis suggests that when modelling industrial energy demand there is a place for ‘endogenous’ technical progress and an ‘exogenous’ underlying energy demand trend; consequently, it is argued that, any modelling strategy should start by including both and only imposing restrictions if accepted by the data.