State space models, Kalman recursions and Bayesian filter

Abstract

The aim of this work was to employ the state space model to Classical and Bayesian estimation and filtering techniques to solve the state estimation problem. The two approaches were tested at the contrast of one another with Kalman recursions on the Classical part and Bayesian filtering on the Bayesian part. Both filters studied the state of the unemployment percentage rates using the local level model. Confidence intervals were computed to enhance the reliability of the estimates and forecasts obtained through the Bayesian filter. The method Maximum Liklihood Estimation and Gauss method of least squares estimates were used to compute the initial parameter estimates required by both filters. A simulation study was undermined to analyze the convergence of the initial parameter estimates required by both filters.