Assessing the performance of nuclear norm-based matrix completion methods on CO2 emissions data

With the aim of performing a counterfactual analysis using the statistical learning method of Matrix Completion (MC) to evaluate the impact of the Emissions Trading System (ETS) - an instrument launched by the European Union in 2005 to reduce CO2 emissions and mit- igate global warming - in this paper we study the performance of recently proposed nuclear norm regularized MC methods for panel data when applied to CO2 emission data. Results show that the inclusion of individual and time fixed effects in the MC optimization problem, and the pre-processing of the original data, increase the performance of the method.