Counterfactual Impact Evaluation of Cohesion Policy 2014-2020: Regression Discontinuity Design

This paper evaluates EU Cohesion Policy interventions during 2014-2020 using regression discontinuity design (RDD), exploiting funding thresholds at 75 and 90 percent of EU average GDP per capita across 281 NUTS-2 regions. The analysis examines nominal and real GDP per capita growth, GDP per capita at purchasing power standards, and employment and unemployment changes. Contrary to earlier programming periods, the study finds no statistically significant discontinuity in treatment intensity near the cutoff thresholds under conventional RDD specifications with flexible functional forms and optimal bandwidth estimators. While marginally significant positive effects on real GDP growth emerge under restrictive assumptions, results lack robustness across model specifications. Pretests reveal potential structural breaks in other causal variables at the cutoffs. The findings provide descriptive evidence of positive correlations between regional development and cohesion policy but fail to establish robust causal links. This limitation likely stems from institutional changes during 2014-2020, including transition region introduction and modified allocation criteria that reduced funding discontinuity sharpness, or from pandemic-driven regional shocks.