HANK on Speed : Robust Nonlinear Solutions using Automatic Differentiation

Building on automatic differentiation, I propose a solution method for heterogeneous agents models with many aggregate equations which allows to account even for strong nonlinearities. A powerful open source reference implementation is provided which typically solves the canonical HANK model within a few seconds, including the nonlinear transition dynamics of the complete distribution. I study a permanent shift in redistribution policy in a medium-scale two-asset HANK model featuring many aggregate frictions. Since firms wish to deplete their capital stock, the transition path is characterized by a long-lasting deflationary episode, which may be intensified by the interest rate lower bound