This study examines the association between natural catastrophes and contract-level default hazard in Italy. We combine granular credit contract data from Experian Italia S.p.A. with EM-DAT disaster records matched at the province-month level over 2000–2025. Default is defined as the first transition from performing to non-performing status, identified by three or more missed installments. We estimate semi-parametric Cox proportional hazards models with time-varying catastrophe exposure, province and sector stratification, interaction terms, catastrophe-time indicators, and robustness controls for calendar-year and month-of-year effects. The results show no robust increase in default hazard during the catastrophe month or the first six months thereafter. Higher hazard ratios emerge in selected later windows, most consistently between seven and twelve months after the event. Sectoral characteristics explain baseline heterogeneity, whereas sector-specific differences in the persistent catas- trophe component are limited. Territorial heterogeneity is more pronounced. The findings highlight delayed and geographically differentiated credit-risk patterns following natural catastrophes.
Natural catastrophes and the dynamics of default hazard: evidence from contract-level credit data / Castellano, R., Cini, F., Mariani, G., Storani, S.. - In: ANNALS OF OPERATIONS RESEARCH. - ISSN 0254-5330. - (2026). [10.1007/s10479-026-07422-4]
Natural catastrophes and the dynamics of default hazard: evidence from contract-level credit data
Rosella Castellano
;Federico Cini;Saverio Storani
2026
Abstract
This study examines the association between natural catastrophes and contract-level default hazard in Italy. We combine granular credit contract data from Experian Italia S.p.A. with EM-DAT disaster records matched at the province-month level over 2000–2025. Default is defined as the first transition from performing to non-performing status, identified by three or more missed installments. We estimate semi-parametric Cox proportional hazards models with time-varying catastrophe exposure, province and sector stratification, interaction terms, catastrophe-time indicators, and robustness controls for calendar-year and month-of-year effects. The results show no robust increase in default hazard during the catastrophe month or the first six months thereafter. Higher hazard ratios emerge in selected later windows, most consistently between seven and twelve months after the event. Sectoral characteristics explain baseline heterogeneity, whereas sector-specific differences in the persistent catas- trophe component are limited. Territorial heterogeneity is more pronounced. The findings highlight delayed and geographically differentiated credit-risk patterns following natural catastrophes.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


