Identifying time-varying and country-specific drivers of sovereign debt risk from credit rating reports

Identifying the drivers of sovereign debt risk is essential for international investment and risk management. However, these drivers exhibit significant country-specific and time-varying heterogeneity. The traditional hypothesis-testing research paradigm faces limitations due to its reliance on limited statistical data and predefined driver sets, which prevents it from accurately capturing such complex dynamics. To address this gap, this study proposes a novel approach that employs text-mining algorithms to extract key sovereign debt risk drivers directly from credit rating reports. Through an empirical analysis of 1190 Moody’s sovereign credit rating reports from 1996 to 2024, we identify eight key drivers. The proposed method captures aggregate-level drivers while systematically accounting for their country-specific and time-varying variations. Drivers characterized by such heterogeneity provide more substantial explanatory and predictive power for sovereign credit ratings and default swap spreads than a generic set of risk factors. These findings are pivotal for enhancing the precision of sovereign debt risk analysis and forecasting.