Identification and Estimation of Heterogeneous Agent Models: A Likelihood Approach
43 Pages Posted: 5 Dec 2017
Date Written: November 08, 2017
In this paper, we study the statistical properties of heterogeneous agent models with incomplete markets. Using a Bewley-Hugget-Aiyagari model we compute the equilibrium density function of wealth and show how it can be used for likelihood inference. We investigate the identifiability of the model parameters based on data representing a large cross-section of individual wealth. We also study the finite sample properties of the maximum likelihood estimator using Monte Carlo experiments. Our results suggest that while the parameters related to the household’s preferences can be correctly identified and accurately estimated, the parameters associated with the supply side of the economy cannot be separately identified leading to inferential problems that persist even in large samples. In the presence of partially identification problems, we show that an empirical strategy based on fixing the value of one the troublesome parameters allows us to pin down the other unidentified parameter without compromising the estimation of the remaining parameters of the model. An empirical illustration of our maximum likelihood framework using the 2013 SCF data for the U.S. confirms the results from our identification experiments.
Keywords: heterogeneous agent models, continuous-time, Fokker-Planck equations, identification, maximum likelihood
JEL Classification: C100, C130, C630, E210, E240
Suggested Citation: Suggested Citation