A Model‐Adaptive Test for Parametric Single‐Index Time Series Models. (28th August 2017)

• Record Type: Journal Article
• Authors: Xia, Qiang; He, Kejun; Niu, Cuizhen
• Is Part Of: Journal of time series analysis. Volume 38:Number 6(2017:Nov.)
• Page Start: 981
• Page End: 999
• Publication Date: 2017-08-28
• Subjects: residual‐marked empirical process -- dimension reduction -- globally smoothing; Time-series analysis -- Periodicals; 519.232
• DOI: 10.1111/jtsa.12249
• Languages: English
• ISSNs: 0143-9782

Abstract

Abstract : In this paper, based on certain residual‐marked empirical processes, we study the model test to validate the composite structure with a given link function for parametric single‐index time series models. To extend an existing directional test that avoids the curse of dimensionality to an omnibus test, a model‐adaptive dimension‐reduction test procedure is proposed. Moreover, to fully utilize the dimension‐reduction structure under the null hypothesis, the test is designed for adapting both the null and alternative hypotheses, which can improve the power for a more general alternative. Simulation results and a real data example show that the proposed method can perform effectively in checking parametric single‐index time series models.