Counterfactual Impact Evaluation of Cohesion Policy 2014-2020: Impact on Enterprises

This paper examines the impact of the European Union's Cohesion Policy 2014-2020 on enterprise dynamics at the NUTS-2 level. Using discrete eligibility thresholds at 75 and 90 percent of EU average GDP per capita, we implement sharp and fuzzy regression discontinuity designs to assess effects on enterprise births and deaths, changes in the number of and employment in enterprises and local units. The analysis draws on ARDECO, DG REGIO, and Eurostat data, and considers both the full period (2014-2020) and a prepandemic subsample (2014-2019). We find no robust evidence of statistically significant discontinuities in treatment intensity at the thresholds, except under restrictive model assumptions. This lack of sharp jumps in funding intensity, combined with low statistical power, prevents credible identification of causal effects on enterprise outcomes. Moreover, diagnostic tests reveal structural breaks in key regional characteristics (e.g., sectoral structure, education, initial enterprise density) at the cutoffs, violating core RDD assumptions and suggesting confounding. We argue that institutional changes – the introduction of "transition regions" category, smoothed eligibility rules, and additional allocation criteria such as unemployment – have weakened the quasi-experimental nature of GDP-based thresholds. Future evaluations should rely on multiperiod designs and alternative identification strategies.