Building Open-Source Empirical Models to Forecast Carbon Emissions

Formulating and implementing climate policy requires a detailed understanding of likely pathways of future carbon emissions. However, most existing tools to provide forecasts for such emissions exhibit several limitations. Projections are predominantly made for the long-term, lacking the necessary detail within the typical policy horizon of one to five years. Moreover, most short-term forecasting models are closed-source ‘black-boxes’ that do not reveal in detail how the obtained forecasts are generated. This paper contributes to closing this gap of a lack of transparent short-term emission forecasts by developing a generalised framework for modelling climate and environmental policies within an empirical macro-econometric modelling approach that is standardised for a large number of countries but allows for immense flexibility. The Aggregate Model is an open-source econometric model builder that primarily aims to provide empirical forecasts of sectoral carbon emissions. The estimated models are based on econometric tools from the robust time series modelling literature such as diagnostic testing, model selection, indicator saturation, and automatic forecast evaluation. We present an illustrative example of how the Aggregate Model as a model-builder tool can be used to forecast different sectoral emissions in Austria. We therefore provide the first short term sectoral emission forecasts for Austria. By putting open-source transparency and replicability at the heart of this project, we contribute to opening up the existing forecasting space to wider academic discussion, criticism, and improvement and hope to provide crucial tools to policy-makers.