Alternative Methods for Solving Heterogeneous Firm Models

This paper implements and compares four alternative techniques for the solution of heterogeneous agents business cycle models within the lumpy capital adjustment framework. The widespread Krusell Smith algorithm consistently delivers high accuracy and economic implications quantitatively similar to other bounded rationality, projection-based approaches, but it does so at the cost of high computational intensity. The Parametrized Distributions and Explicit Aggregation methods yield important speed gains but reduced accuracy. The conceptually distinct Projection plus Perturbation method implies qualitatively similar economic results, even more dramatic reductions in computational cost, as well as an important scalability of the aggregate state space. A code package implementing each solution method is available online.