Autore
Gil-Alana, Luis A.

Titolo
Nelson and plosser revisited: Evidence from fractional ARIMA models
Periodico
European University Institute of Badia Fiesolana (Fi). Department of Economics - Working papers
Anno: 1998 - Fascicolo: 21 - Pagina iniziale: 1 - Pagina finale: 35

Fractionally integrated ARMA (ARFIMA) models are investigated in an extended version of Nelson and Plosser's (1982) data set. The analysis is carried out using Sowell's (1992) procedure of estimating by maximum likelihood in the time domain. A crucial fact when estimating with parametric approaches is that the model must be correctly specified. Otherwise, the estimates are liable to be inconsistent. A model-selection procedure based on diagnostic tests on the residuals, along with likelihood criterions is adopted to determine the correct specification of each series. The results suggest that all series except unemployment rate and bond yield are integrated of order greater than one. Thus, the standard approach of taking first differences may lead to stationary series with long memory behaviour.




Esportazione dati in Refworks (solo per utenti abilitati)