Convergence through sustainable development: can EU developing regions make it happen? firm-level counterfactual evidence via Machine Learning

This work investigates whether EU cohesion policies aiming at environmental improvement and carbon reduction have an economic impact on adopters. By merging data from Opencoesione, with firms’ information from AIDA, we look at the changes in firms’ performance due to the sustainability-oriented technologies financed by the European cohesion funds during the 2007–13 programming period. We include firms that participated in pilot programs and received public incentives to upgrade their production plants with sustainable technologies, and we use Machine Learning (ML) techniques to identify the most appropriate counterfactuals for a multilevel DiD setting. Our results indicate a strong and positive policy effect on firms’ profitability, with dissimilar dynamics for different levels of public support. Additionally, over time, the policy effect on treated firms tends to diminish, suggesting the possibility of a rebound effect where the gains in production efficiency and energy savings may be, at least partially, repurposed by firms to increase production (and profits) instead of reducing absolute emissions. This perfectly aligns with what one can expect from economic agents at the micro-level: firms’ actions are guided by the search for ways to obtain profit increases. However, at the macro-level, policymakers should question if the policy design could be improved through the adoption of conditional subsidies or regulatory mechanisms that, by limiting emissions, could foster more environmental benefits.