A Quantitative Theory of Hard and Soft Sovereign Defaults

Empirical research on sovereign default shows "hard defaults"---defined as defaults with above-average haircuts---have worse outcomes for GDP growth than "soft defaults" and that sovereigns continue to borrow post-default. We propose a model capable of capturing these and other empirical regularities. In it, the sovereign makes period-by-period decisions of whether to make the prescribed debt payments or not. Hard defaults arise when the sovereign repeatedly \emph{chooses} to not pay over the course of many years. Unlike in the standard model, default does not exogenously result in autarky. Rather, autarky-like conditions arise endogenously as the shocks leading to default result in higher spreads than the sovereign is willing to pay. The calibrated model predicts that growth shocks are the main determinant of whether default is hard or soft. We use the model and the particle filter to decompose how much of the empirical correlation between default intensity and output growth is selection and how much is causal. Decomposition of model forces shows that one-third (one-tenth) of hard (soft) defaults are explained by actual default costs with the rest explained by selection. A historical decomposition of shocks reveals that transitory shocks and trend shocks were the primary drivers of the Argentinean defaults in the 1980s and the 2000s, respectively. Estimated haircuts were 20 percentage points higher in the 2001 default than in the one in the 1980s, consistent with the data. Our estimated productivity shocks coincide with major events such as the convertibility plan and the Asian crisis.