One of the most challenging aspects of (environmental) policy impact evaluation studies is the construction of credible counterfactuals for causal inference. In this paper, we leverage a method that does not require the creation of benchmarks or the identification of external comparable control units to evaluate the effectiveness of the first three phases of the European Union Emissions Trading System (EU ETS) (2005-2019) in reducing power sector fossil fuel CO2 emissions across 24 EU ETS Member States. Considering the beginning of each trading period (phase) as a policy intervention, the paper adopts a Bayesian structural time series (BSTS) modeling framework alongside a set of contemporaneous predictors related to power sector emissions to build counterfactual estimates of emissions for each post-intervention period and analyze the policy implementation effect by comparing actual emissions with counterfactual estimates. The results indicate a statistically significant emissions reduction in the second and third phases. The dominance of the power sector within the ETS since its inception emphasizes the importance of our findings in advancing emissions reduction objectives.
Keywords: The EU ETS, Fossil Fuel CO2 Emissions, Climate Variables,Predictive Modeling, Spatio-temporal Analysis, Counterfactual Inference.
