Estimating Heterogeneous Economies with Micro Data

Abstract

We give sufficient conditions under which dynamic equilibrium models with heterogeneous-agents can be represented by a first-order reduced-rank vector autoregression. We exploit this result to develop an econometric framework that enables the rapid estimation of a rich class of models with macro and repeated cross-sections of micro data. Simulation evidences suggest that our method delivers increased precision of parameter estimates than conventional approaches, particularly for parameters governing cross-sectional dynamics. We apply our method to estimate a medium-scale HANK model with heterogeneous earnings exposures to aggregate fluctuations at the household-level. Our estimates imply that low-income households are more sensitive to changes in aggregate income on average, and that this sensitivity is heightened conditional on monetary policy shocks. Through the lens of the model, our method estimates that heterogeneous earnings exposures amplify the aggregate consumption response to monetary policy shocks by 40%, substantially larger than those implied by traditional estimation methods.