Regional and Sectoral Heterogeneity in Corporate Insolvency: Evidence for Spain Using Fixed Effects Models and Machine Learning

This study examines the macro-financial determinants of corporate insolvency dynamics in Spain over the period 2008–2024 using a region–sector panel of annual data. The analysis focuses on how systemic financial conditions, macroeconomic fundamentals, and territorial heterogeneity jointly shape regional patterns of business failure. Financial indicators are condensed through principal component analysis into two latent factors capturing complementary dimensions of risk: a domestic financial stress component associated with equity and banking volatility, and a global uncertainty component linked to exchange-rate movements and external shocks. Panel estimations reveal that domestic financial pressure constitutes the main transmission channel of corporate fragility across regions, while global risk operates through asymmetric and non-linear mechanisms. Macroeconomic conditions, particularly unemployment dynamics, reinforce these effects, whereas inflation shows limited direct influence on insolvency outcomes.

To assess robustness and predictive relevance, the econometric framework is complemented by machine learning models including CatBoost, XGBoost, and LightGBM. The ensemble approach achieves high and stable predictive performance, with AUC values above 0.90, confirming that macro-financial variables contain persistent information about regional failure risk. Clustering analysis further identifies distinct behavioural profiles across regions and sectors, highlighting heterogeneous exposure to systemic shocks. Overall, the results demonstrate that corporate insolvency in Spain is primarily driven by the interaction between financial-market volatility and regional economic structures. The proposed framework contributes to the development of macro-financial early-warning systems and provides new evidence on how systemic risk propagates spatially within national economies.