MAXIMUM LIKELIHOOD ESTIMATION OF PARAMETERS IN STATE-SPACE MODELS

摘要

This paper presents the possibility of likelihood function computation for state-space econometric models. Estimation outputs using the Kalman filter are utilized to compute the likelihood function of observed data sequences. The likelihood function is then utilized for parameter estimation. The advantage of this method is its relative facility and especially the possibility to use this method for nonlinear state-space models too. Other advantages are seen in the initial conditions estimation of transition economy models, which requires time-variant parameters, and in the case of initial conditions estimation, when thee are unobservable variables in the model.