Forecasting Time Series Subject to Multiple Structural Breaks

• Published in: The Review of economic studies Vol. 73; no. 4; pp. 1057 - 1084
• Main Authors: Pesaran, M. Hashem, Davide Pettenuzzo, Timmermann, Allan
• Format: Journal Article
• Language: English
• Published: Oxford Wiley-Blackwell 01.10.2006; Review of Economic Studies Ltd; Oxford University Press
• Subjects: Bayesian analysis; Bayesian method; C22; C53; E43; E47; Econometric models; Economic forecasting models; Economic forecasts; Economic models; Estimation; Estimators for the mean; Forecasting; Forecasting models; Forecasting techniques; Government bonds; Markov analysis; Model testing; Modeling; Parametric models; Predictive modeling; Probability forecasts; Rates of return; Statistical variance; Stochastic models; Structural change; Studies; Time series; Time series forecasting; Treasury bills; U.S.A; United States > US
• ISSN: 0034-6527; 1467-937X
• DOI: 10.1111/j.1467-937X.2006.00408.x

This paper provides a new approach to forecasting time series that are subject to discrete structural breaks. We propose a Bayesian estimation and prediction procedure that allows for the possibility of new breaks occurring over the forecast horizon, taking account of the size and duration of past breaks (if any) by means of a hierarchical hidden Markov chain model. Predictions are formed by integrating over the parameters from the meta-distribution that characterizes the stochastic break-point process. In an application to U.S. Treasury bill rates, we find that the method leads to better out-of-sample forecasts than a range of alternative methods.