Empirical Assessment of the Multiplier Effects of the EU Recovery and Resilience Facility Using Machine Learning

This research demonstrates the potential of machine learning for revealing the fiscal effects of the Recovery and Resilience Facility (RRF) in the European Union. It focuses on the use of a hybrid approach, based on traditional econometric methods combined with advanced data machine learning techniques. For this purpose, the following were applied: a panel data model with fixed effects, difference-in-differences analysis, correlation analysis, and machine learning, specifically, random forest regression, for the period of 2020–2024, including indicators from all 27 EU member states. The results of the conducted tests establish the effectiveness of the Recovery and Resilience Facility for fiscal stabilization, but also its high vulnerability to specific economic conditions in the individual member states. The complex relationships between the amount of funds received and the fiscal outcomes, which classical models fail to capture, were derived. A positive stabilizing effect on the indebtedness of countries with a clearly expressed imbalance in the public debt-to-gross domestic product ratio was demonstrated.