Statistics Seminar（2017-02）

Topic:Gradient-Based Structural Change Detection For Non-Stationary Time Series M-estimation

Speaker:Weichi Wu, University College London

Time:Tuesday, 10 January,10:00-11:00

Place:Room 216, Guanghua Building 2

Abstract:

We consider structural change testing for a wide class of time series M-estimation with non-stationary predictors and errors. Flexible predictor-error relationships, including exogenous, state-heteroscedastic and autoregressive regressions and their mixtures, are allowed. New uniform Bahadur representations are established with nearly optimal approximation rates. A CUSUM-type test statistic based on the gradient vectors of the regression is considered. In this paper, a simple bootstrap method is proposed and is proved to be consistent for M-estimation structural change detection under both abrupt and smooth non-stationarity and temporal dependence. Our bootstrap procedure is shown to have certain asymptotically optimal properties in terms of accuracy and power. A public health time series data set is used to illustrate our methodology, and asymmetry of structural changes in high and low quantiles are found.

Introduction:

Weichi Wu is a research associate at Department of Statistical Science and Big Data Institute, University College London. His research interests include non-stationary times series, network data analysis, non-parametric method, change point problem and M-estimation. His research has been published in Journal of Business & Economic Statistics. He earned his Ph.D in statistics at University of Toronto, MA. in statistics at Columbia University in the City of New York, and B.S in Physics at Peking University.

Your participation is warmly welcomed!