Small Sample Properties of Forecasts from Autoregressive Models Under Structural Breaks

By M. Pesaran, Allan G. Timmermann. Social Science Research Network, 2003

458 citations recorded by the source corpus.

Autoregressive models are used routinely in forecasting and often lead to better performance than more complicated models. However, empirical evidence is also suggesting that the autoregressive representations of many macroeconomic and financial time series are likely to be subject to structural breaks. This paper develops a theoretical framework for the analysis of small-sample properties of forecasts from general autoregressive models under a structural break. Our approach is quite general and allows for unit roots both pre- and post-break. We derive finite-sample results for the mean squared forecast error of one-step-ahead forecasts, both conditionally and unconditionally and present numerical results for different types of break specifications. Implications of breaks for the determination of the optimal window size are also discussed.

Subfields: Economics and Econometrics

View Paper: https://www.semanticscholar.org/paper/d91e0843ae2ae74628b43cfbc83acf195a744bb1

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Economics and Econometrics

Citations in the corpus

Consistent Factor Estimation in Dynamic Factor Models with Structural Instability (2013, background): “Pursuing this calculation would be of independent interest and would also return the large-dimensional discussion here to the bias-variance tradeoffs associated with ignoring breaks tackled in a low-dimensional setting by Pesaran and Timmermann (2005, 2007).”
The Essential Role of Pandemics: A Fresh Insight Into the Oil Market (2020, unknown): “Additionally, the choice of the rolling-window width cannot be less than 20 in the subsample test (Pesaran & Timmermann, 2005); we, thereby, select 24-months2 to ensure the robustness of the empirical results (Su, Qin, et al., 2019).”
How strong is the linkage between tourism and economic growth in Europe (2015, background): “According to Pesaran and Timmermann (2005), the choice of the window length should be determined by the existence of structural breaks in the series, and the length of the post-break observations.”
General Diagnostic Tests for Cross Section Dependence in Panels (2020, unknown): “(See Pesaran and Timmermann (2004)).”
Model Averaging in Economics: An Overview (2015, unknown): “Raftery (1995), Volinsky et al. (1997), Fernández et al. (2002)), and, on the other hand, papers from the forecasting combination literature surveyed in Section 5.”
Selection of estimation window in the presence of breaks (2007, background): “Autoregressive processes subject to structural breaks have been considered by Pesaran and Timmermann (2005). The advantage of that setup is that it is highly relevant to many empirical situations, but the theoretical results are complicated to interpret. Under strictly exogenous regressors, however,”