Automatic and Open-Source Model with Forecasts for Climate Policy and Economics

Economic forecasts of the effects of climate policy are frequently based on static economic theory and are not regularly updated. Moreover, their source code is often not public, making replication and critical evaluation difficult. The predominant models used for climate policy narratives, so called Integrated Assessment Models, are rarely estimated using empirical data and are hence highly affected by the modeller’s assumptions. To improve the estimation of likely effects of climate policy, we present the “Aggregate Model”, a data-based model for dozens of countries that flexibly estimates and forecasts economic time series and allows for the simulation of different climate policy options. Data-based models need to incorporate long-term trends and account for both structural breaks and outliers that otherwise distort the model estimates and may lead to systematic forecast error. Our model uses various techniques from the time series literature, such as indicator saturation, model selection, and testing for co-integration. These techniques are automated to a high degree, simplifying the model estimation procedure.  The Aggregate Model is distributed as an open-source R package, allowing for simple replication and modification by users through its modular structure.