Predicting local taxation decision-making: evidence from Italian municipalities

This study investigates whether local taxation decisions can be reliably predicted using machine learning (ML), focusing on the personal income tax sur charge set by Italian municipalities. We evaluate the predictive performance of ML models and identify key determinants of tax-setting behaviour. Results show that municipal tax choices follow systematic and predictable patterns driven by demographic, fiscal, socio-economic and institutional factors. A North–South comparison reveals stronger predictive accuracy in Southern regions, suggesting that tax increases are more closely linked to fiscal constraints than to discretionary political choices. Our findings highlight the usefulness of ML for understanding and supporting local fiscal planning.