Titolo della tesi: Balancing Score Methods for Continuous Treatment Evaluation: Evidence from the Common Agricultural Policy
This dissertation develops an applied framework for causal inference with continuous treatments using balancing score-based methods. It first provides a methodological guide to generalized propensity score estimation, recent advances in covariate balancing and machine learning approaches, and the associated diagnostics for overlap and balance. The empirical application estimates the dose–response function of the Common Agricultural Policy (CAP) Pillar I support on regional economic and employment outcomes across European regions (2011–2015). Results reveal heterogeneous and non-linear effects: Pillar I support enhances agricultural and total gross value added and mitigates agricultural employment decline, with the strongest effects emerging at intermediate treatment intensities. These findings highlight both the methodological value of continuous-treatment evaluation and the dual economic role of CAP Pillar I ---as a stabiliser of agricultural incomes and a facilitator of regional adjustment.