No Detail Found!

On the performance of wild bootstrap based on mm-gm6 estimator in the presence of heteroscedastic errors and high leverage points

Author(s)
Osama, A.A. Alsattari Al-Aqsa University, Khan Younis, Gaza strip, Palestine
Paul, I. Dalatu Adamawa State University, Mubi-Nigeria
Abstract
The violation of constancy of variance of error terms causes the problem of heteroscedasticity. Even though the ordinary least squares (OLS) estimates are unbiased in the existence of heteroscedasticity problem in a data set, the standard errors of the parameter estimates are biased. This renders the estimator inefficient. As an alternative, a weighted residual (wild bootstrap) may be used to remedy this problem. However, the weakness of wild bootstrap is that, in the presence of outliers, the estimates of the standard errors become large. For the sake of rectifying this problem, a wild bootstrap (WB) based on MM estimates is proposed. Nevertheless, this estimator cannot handle well high leverage points (HLPS). Thus, wild bootstrap based on MM-GM6 estimator is proposed so that the problems of both heteroscedasticity and outliers can be rectified. The performance of the proposed method denoted as WBootMM-GM6-Liu is compared with some existing techniques such as wild bootstrap of OLS (WBootOLS), wild bootstrap of Liu (WBootLiu) and wild bootstrap based on MM estimator denoted as (WBootMM-Liu). The numerical results indicate that the developed method outperformed other methods for data having both problems of heteroscedasticity and high leverage points.
Publication Details
Page(s) 121-134
DOI DOI not available
Published Journal: Pakistan Journal of Statistics, Volume: 37, Issue: 2, Year: 2021

Request the full text from PASTIC

Keywords
outliers, Heteroscedasticity, MM-GM6 estimator, weighted residual, wild bootstrap
0

Citations

0

Downloads

70

Views