The heterogeneous causal effects of the EU's Cohesion Fund

Abstract: This paper quantifies the causal effect of cohesion policy on EU regional output and investment focusing on one of its least studied instruments, i.e., the Cohesion Fund (CF). We employ modern causal inference methods to estimate not only the local average treatment effect but also its time-varying and heterogeneous effects across regions. Utilizing this method, we propose a novel framework for evaluating the effectiveness of CF as an EU cohesion policy tool. Specifically, we estimate the time varying distribution of the CF's causal effects across EU regions and derive key distribution metrics useful for policy evaluation. Our analysis shows that relying solely on average treatment effects masks significant heterogeneity and can lead to misleading conclusions about the effectiveness of the EU's cohesion policy. We find that the impact of the CF is frontloaded, peaking within the first seven years after a region's initial inclusion in the program. The distribution of the effects during this first seven-year cycle of funding is right skewed with relatively thick tails. This indicates positive effects but unevenly distributed across regions. Moreover, the magnitude of the CF effect is inversely related to a region's relative position in the initial distribution of output, i.e., relatively poorer recipient regions experience higher effects compared to relatively richer regions. Finally, we find a non-linear relationship with diminishing returns, whereby the impact of CF declines as the ratio of CF funds received to a region's gross value added (GVA) increases.