Empirical analysis of monetization and sovereign default: a logit and random forest analysis

There is a prevailing claim, particularly among proponents of modern monetary theory (MMT), which sovereign defaults on public debt denominated in a sovereign currency can be avoided through monetization. Despite considerable opposition to this assertion, to the best of the authors‘ knowledge, no rigorous empirical analysis has been conducted to test this proposition. This study aims to fill this gap by statistically analysing the validity of the proposition using traditional logistic regression and a machine-learning technique known as random forest. The findings suggest that, although the share of public debt denominated in a sovereign currency indeed reduces the probability of sovereign default, other factors, such as the public debt-to-gross domestic product ratio and the economic growth rate, do significantly influence the probability of sovereign default. These results indicate that the data do not support the MMT’s assertion.