Sankhya: The Indian Journal of Statistics

2005, Volume 67, Pt. 2, 200--226

Bootstrap in Detection of Changes in Linear Regression


Marie Hu\v{s}kov\'a, Charles University in Prague, and Czech Academy of Sciences, Czech Republic
Jan Picek, Technical University of Liberec, Czech Republic

SUMMARY. Applications of bootstrap with and without replacement in  change point analysis in linear regression models are discussed. Particularly, bootstrap based approximations for critical values for two classes of $M$-type test procedures are treated.  As a particular case, we obtain $L_1$ procedures and regression quantile procedures. Their asymptotic  performance is investigated and finite sample properties are checked in a~simulation study.

AMS (1991) subject classification. 62G20, 62E20, 60F17.

Key words and phrases. Linear regression, structural changes, $M$-test procedures, bootstrap with and without replacement.

Full paper (PDF)