Accelerating the resolution of sovereign debt models using an endogenous grid method

State-of-the-art algorithms for solving sovereign debt models with endogenous default rely on value function iterations. These algorithms are consequently very slow and quickly become intractable, even for a state space of dimension as low as three. This paper shows how to adapt the endogenous grid method for sovereign debt models, leading to a dramatic speed gain by a factor comprised between 5 and 10. A second contribution is to quantify and compare the accuracy of the computed solutions by both the value function iterations and the endogenous grid method.