INVESTIGATING THE DETERMINANTS OF LOCAL INDEBTEDNESS IN SLOVAKIA: THE MACHINE LEARNING APPROACH

The paper aims to investigate local indebtedness determinants in a sample of Slovak municipalities using machine learning methods based on decision trees. The sample covers all 2,926 municipalities listed by the Ministry of Finance of the Slovak Republic in 2005 – 2022. Our results point to significant effect of current expenditure, subsidies, size category, and crises on local indebtedness using the QUEST, CRT, CHAID, and exhaustive CHAID growing methods. Based on the results of decision trees, significant variables are treated as regressors in logit and probit econometric estimations, to check their statistical significance. The results of the estimations correspond to the results of decision trees and provide us with a further view of the determinants of local indebtedness in Slovakia.