Modelos de memoria larga para series económicas y financieras

This paper provides a review of time series models with long memory in the mean and conditional variance, with special attention to Fractionally Integrated ARMA processes (ARFIMA) and fractionally integrated GARCH and SV processes. Their more important properties are reviewed and its application to model economic and .nancial time series is discussed. The main estimation methods and tests proposed in the literature for the long memory property are also reviewed. Finally, this paper reviews the main results on prediction of future values of long memory time series.