Estimating Nonlinear Heterogeneous Agents Models with Neural Networks
Working Papers
Abstract
We leverage the recent advancements in machine learning to develop a solution and estimation method based on neural networks for complex economic models. We apply our method to a nonlinear Heterogeneous Agent New Keynesian (HANK) model with a zero lower bound constraint for the nominal interest rate. To begin with, we demonstrate with simulated data that our method is much more efficient than existing global solution methods and that likelihood estimation converges to the true parameter values. We then estimate the model with US aggregate data and evaluate the information content it provides about households’ idiosyncratic risk. Our empirical application also sheds light on the efficacy of our method in simultaneously handling a large number of state variables and parameters, nonlinear dynamics, heterogeneity, and aggregate uncertainty.