Representación en el espacio de los estados y filtro de Kalman en el contexto de las series temporales económicas

Abstract

The key to handling structural time series models is the state space form. The importance of the state space model for the statistical analysis of time series is primarily due to the Kalman filter and its relation to the prediction error decomposition which allows the likelihood function of state space models to be evaluated in a simple and straightforward way. Another feature of the state space framework is its generality: most practical gaussian time series models can be formulated as a state space model. The goal of this paper is to describe the state space form, the Kalman filter and smoothing algorithms and the maximun likelihood estimation in the time domain. Also, the role of inital conditions and outliers for the Kalman filter is explored. Finally, the procedure is applied to the most important structural models.