Full-Information Estimation of Heterogeneous Agent Models Using Macro and Micro Data

We develop a generally applicable full-information inference method for heterogeneous agent models, combining aggregate time series data and repeated cross sections of micro data. To handle unobserved state variables that affect cross-sectional distributions, we compute numerically unbiased estimate the model-implied likelihood function. Employing in Markov Chain Monte Carlo algorithm, obtain fully efficient valid Bayesian inference. Evaluation part lends itself naturally to parallel computing. Numerical illustrations models with households or firms demonstrate proposed substantially sharpens relative using only macro data, some parameters is essential identification.