Efficient Likelihood Evaluation of State-Space Representations

Abstract

We develop a numerical procedure that facilitates efficient likelihood evaluation in applications involving non-linear and non-Gaussian state-space models. The procedure employs continuous approximations of filtering densities, and delivers unconditionally optimal global approximations of targeted integrands to achieve likelihood approximation. Optimized approximations of targeted integrands are constructed via efficient importance sampling. Resulting likelihood approximations are continuous functions of model parameters, greatly enhancing parameter estimation. We illustrate our procedure in applications to dynamic stochastic general equilibrium models.