How Much Is Too Much? Assessing the Nonlinear Relationship Between Debt and Sovereign Creditworthiness

This paper explores the rich relationship between economic fundamentals and sovereign creditworthiness by using statistical learning approaches to explicitly model the debt thresholds associated with rating changes. Crucially, less structure is imposed than by standard linear models. The endogenous structure emerging from the data is consistent with economic theory and confirms the existence of profound non-linearities. In particular, while an increase in public debt reduces creditworthiness, the extent crucially depends on a country’s specific situation. Imposing less structure also provides additional insights, for instance, that for countries with a higher GDP per capita, a change in debt levels is more likely to result in a rating change, and that certain ratings (e.g., Baa) are associated with a larger range of fundamentals. Extensive robustness tests confirm the results, which may also be relevant for other types of borrowers.