Researchers have not reached a consensus on how to account for technic
al progress when modeling aggregate energy demand. Some include a time
trend in a dynamic model, while others omit the trend in a model that
disallows for long-run income effects. We show that the latter approa
ch can be directly tested as a nested case of the former, so that mode
l selection need not be made a priori, solely on theoretical grounds.
Using annual data over 1960-90, we find that long-run income effects a
re not significant; however, including a time trend improves the model
's fit while rendering much more credible long-run price elasticities.