Solving HACT models with bankruptcy choice

We introduce bankruptcy choice to the heterogeneous agent in continuous time (HACT) framework developed in Achdou et al. (2022). We demonstrate that real-options-like problems such as the decision to declare bankruptcy can be efficiently solved using the ``value-matching'' condition only (unlike alternative methods that require both value matching and ``smooth pasting''). Moreover, we show that under certain conditions, smooth-pasting may not hold. Given this, we recommend (and demonstrate the use of) linear complementarity problem (LCP) solvers for real-option like problems, especially in settings where control variables depend on the slope of the value function. We show that this approach is more flexible and computationally efficient than other popular solution methods. In particular, it is less prone to errors in settings that have corner solutions.