Journal of Statistics: Advances in Theory and ApplicationsAuthor's: Shipra Banik and B. M. Golam Kibria
Pages: [1] - [21]
Received Date: October 21, 2018
Submitted by:
DOI: http://dx.doi.org/10.18642/jsata_7100122013
This paper considers some bootstrap version of the existing confidence intervals for estimating the parameter of an autoregressive process of order one model. A simulation study has been conducted to compare the performance of the proposed intervals using two important measures: coverage probability and average width. It appears from our simulation study that all methods have confidence coefficient closest to the given confidence coefficient, however, our proposed bootstrap intervals have small average widths as compare to its counterpart. A real life data are analyzed, which supported the simulation results to some extent. We believe that the findings of this study will make important contribution to the time series literature.
bootstrapping, confidence intervals, simulation, stock price index, time series models.
