A statistical model to forecast and simulate energy demand in the long-run
Ignacio Mauleón
Abstract
This research aims to design a model to forecast and simulate aggregated world energy demand at distant horizons in time. This is done by estimating statistically a simplified interrelated model for the three variables considered, total primary Energy, world population and GDP. The approach intends to offer a complementary perspective to current practice, based on simulating energy demand conditional on GDP and population. The model is based on long historical series spanning the years (1900;2017) available from renowned researchers and institutions in their respective fields. The estimated models allow a forecast of future energy demand and a risk/sensitivity analysis. Alternative solutions and simulation methods are carried out to assess the robustness of the results derived. These forecasts are compared to the results of key relevant roadmaps put forward in the literature - in the range (330;408) EJ/yr for final energy consumption -, the general conclusion being that the aforementioned roadmaps assume sizeable efficiency savings, relying mainly on electrification and renewable energies deployment, that depart significantly from historical trends embodied in the model estimated - 900 EJ/yr on average, and 600 EJ/yr under favourable assumptions. These results jeopardise the unbounded GDP-growth paradigm, suggesting a replacement by alternative welfare measures as suggested by the UN human development index, the prosperity approach, and related standards.