Validating Allocation Functions in Energy Models: An Experimental Methodology

In the late 1970s, the Energy Information Administration initiated a program calling for review and evaluation of its data and validation of energy models used in support of the policymaking process. One of the more controversial aspects of this program was the effort to validate the large-scale energy models developed under DOE auspices and, in some cases, still under development. As all participants in this process (i.e., modelers and evaluators) acknowledge, there is no absolute standard by which a model can be validated. By definition, a model is an approximation to some real-world process. It abstracts from the complexities of the process but is intended to capture essential dimensions of the forces governing outcomes of that process. Consequently, all evaluations of a model involve judgment. To illustrate the prospects for divergent yet individually sensible judgments of an energy model, consider a recent appraisal of the Regional Demand Forecasting Model (RDFOR).