Open Access
2019 Predictive, finite-sample model choice for time series under stationarity and non-stationarity
Tobias Kley, Philip Preuß, Piotr Fryzlewicz
Electron. J. Statist. 13(2): 3710-3774 (2019). DOI: 10.1214/19-EJS1606


In statistical research there usually exists a choice between structurally simpler or more complex models. We argue that, even if a more complex, locally stationary time series model were true, then a simple, stationary time series model may be advantageous to work with under parameter uncertainty. We present a new model choice methodology, where one of two competing approaches is chosen based on its empirical, finite-sample performance with respect to prediction, in a manner that ensures interpretability. A rigorous, theoretical analysis of the procedure is provided. As an important side result we prove, for possibly diverging model order, that the localised Yule-Walker estimator is strongly, uniformly consistent under local stationarity. An R package, forecastSNSTS, is provided and used to apply the methodology to financial and meteorological data in empirical examples. We further provide an extensive simulation study and discuss when it is preferable to base forecasts on the more volatile time-varying estimates and when it is advantageous to forecast as if the data were from a stationary process, even though they might not be.


Download Citation

Tobias Kley. Philip Preuß. Piotr Fryzlewicz. "Predictive, finite-sample model choice for time series under stationarity and non-stationarity." Electron. J. Statist. 13 (2) 3710 - 3774, 2019.


Received: 1 January 2019; Published: 2019
First available in Project Euclid: 1 October 2019

zbMATH: 07113729
MathSciNet: MR4013749
Digital Object Identifier: 10.1214/19-EJS1606

Primary: 62M20
Secondary: 62M10

Keywords: covariance stationarity , forecasting , local stationarity , Yule-Walker estimate

Vol.13 • No. 2 • 2019
Back to Top