Author(s):
Gabriel, Vasco J. 
; Martins, Luís F.
Date: 2000
Persistent ID: http://hdl.handle.net/1822/1414
Origin: RepositóriUM - Universidade do Minho
Subject(s): Long memory; Structural change; Forecasting
Document details
The forecast performance of long memory and Markov switching models
Description
Recent research has focused on the links between long memory and structural change, stressing the long memory properties that may arise in models with parameter changes. In this paper, we contribute to this research by comparing the forecasting abilities of long memory and Markov switching models. Two approaches are employed: a Monte Carlo study and an empirical comparison, using the quarterly Consumer Price inflation rate in Portugal in the period 1968-1998. Although long memory models may capture some in-sample features of the data, when shifts occur in the series considered, their forecast performance is relatively poor, when compared with simple linear and Markov switching models. Moreover, our findings, in a more general framework, are in accordance with the works of Clements and Hendry (1998) and Clements and Krolzig (1998), reinforcing the idea that simple linear time series models remain useful tools for prediction.
Recent research has focused on the links between long memory and structural change, stressing the long memory properties that may arise in models with parameter changes. In this paper, we contribute to this research by comparing the forecasting abilities of long memory and Markov switching models. Two approaches are employed: a Monte Carlo study and an empirical comparison, using the quarterly Consumer Price inflation rate in Portugal in the period 1968-1998. Although long memory models may capture some in-sample features of the data, when shifts occur in the series considered, their forecast performance is relatively poor, when compared with simple linear and Markov switching models. Moreover, our findings, in a more general framework, are in accordance with the works of Clements and Hendry (1998) and Clements and Krolzig (1998), reinforcing the idea that simple linear time series models remain useful tools for prediction.
Document Type
Research paper
Language English
Language English
| Financiadores do RCAAP | |||||||
|
|
|
|
|
|
||