Abstract:
The autoregressive conditional heteroscedastic (ARCH) regression models are widely used in various branches of econometrics. In this paper, we focus on a two stage estimation of mean parameters of linear regression model with ARCH errors. In our estimation procedure, we first compute initial consistent estimates and then conduct estimated weighted least squares based on the first-stage estimates. We show that the proposed estimator is more efficient than the quasi-maximum likelihood estimator (QMLE).
Page(s):
261-274
DOI:
DOI not available
Published:
Journal: Pakistan Journal of Statistics, Volume: 23, Issue: 3, Year: 2007