Modelling the unit root properties of electricity data—A general note on time-domain applications

The logic of identifying the stationary features of energy series lays in the policy potentials contained within unit root-related information. By providing an overview of unit root testing frameworks applied on electricity data, this paper underlines why knowledge derived from integration order can benefit to electricity consumption/production forecasting, modelling, and planning. In a methodological discussion, empirical divergences reported in the literature are linked to the variety of unit root testing baselines employed, and the heterogeneous manners through which past stationarity analyses approached identification caveats in time-series econometrics, and threats to internal validity. Further, we argue that time-series model specifications and data choices do matter. They co-determine the external validity of the empirics and may hinder the quality of policy recommendations if mis-considered. Doors for future examinations of stochastic features of electricity series are finally opened, with research opportunities for both theorists and empiricists.