Accuracy Verification for Numerical Solutions of Equilibrium Models

We propose a simulation-based procedure for evaluating approximation accuracy of numerical solutions of general equilibrium models with heterogeneous agents. We measure the approximation accuracy by the magnitude of the loss suered by the agents as a result of following suboptimal policies. Our procedure allows agents to have knowledge of the future paths of the economy under suitably imposed costs of such foresight. This method is very general, straightforward to implement, and can be used in conjunction with various solution algorithms. We illustrate our method in the context of the incomplete-markets model of Krusell and Smith (1998), where we apply it to two widely used approximation techniques: cross-sectional moment truncation and history truncation.