Autori
Maravall, Augustin
Gomez, Victor

Titolo
Time series regression with Arima noise and missing observations program TRAM
Periodico
European University Institute of Badia Fiesolana (Fi). Department of Economics - Working papers
Anno: 1992 - Fascicolo: 81 - Pagina iniziale: 1 - Pagina finale: 158

The present paper describes the program TRAM, which stands for "Tim e Series Regression with ARIMA Noise and Missing Observations". TRAM has been written in Fortran and is available from the authors for MS-DOS computers and mainframes. The program estimates the parameters of a regression model with possibly nonstationary noise and any sequence of missing observations, interpolates the missing values, and obtains forecasts for the series. The program incorporates several additional facilities, such as intervention analysis, and easter and/or trading day corrections. The methodology is described in the paper "Estimation, Prediction and Interpolation for Nonstationary Series with the Kalman Filter", by V. Gomez and A. Maravall, EUI Working Paper ECO No. 92/80. The first part of this document presents a summary of the program. Part two contains the instructions for the user and a description of the parameters. Finally, part three illustrates the program for six well-known examples, which present different regression and time series models with different combinations of missing observations.




Esportazione dati in Refworks (solo per utenti abilitati)