Una nota sobre la estimación eficiente de modelos con parámetros cambiantes

Standard estimation procedures for the time-varying parameters model suppose that the variances of the noises in the model are known. Obviously, this assumption is not realistic in most econometric applications. Besides, the results of these methods are sensitive to the initial conditions of the algorithm, a fact that is often overlooked by the literature. In this paper, we propose an extension of the recursive algorithm proposed by Cooley, Rosenberg y Wall (1977), which is independent of initial conditions and includes on-line estimation of all the relevant variances. The results obtained with this method compare favourably with those obtained by standard procedures.