Public Seminar of PhD Candidate on "Robust diagnostic checking, quantile inference & LAD est..."
posted by Department of Statistics and Actuarial Science for HKU and Public
Event Type: Public Lecture/Forum/Seminar/Workshop/Conference/Symposium
Event Nature: Science & Technology
Public Seminar of PhD Candidate
Ms. ZHENG Yao
Department of Statistics and Actuarial Science
The University of Hong Kong
will give a talk
ROBUST DIAGNOSTIC CHECKING, QUANTILE INFERENCE AND LAD ESTIMATION FOR SOME TIME SERIES MODELS
In this talk, I will discuss the robust diagnostic checking, quantile inference, and least absolute deviations (LAD) estimation for some time series models. Firstly, although there has been extensive research on estimation for time series models with heavy-tailed innovations, the corresponding goodness-of-fit tests are still lacking. Based on the idea of transforming the absolute residuals, I will propose a goodness-of-fit test for the widely used generalized autoregressive conditional heteroscedastic (GARCH) model which is applicable to arbitrary tail-heaviness. Secondly, I will consider the quantile regression for the GARCH model. While this model has a seemingly intractable conditional quantile structure, I will show that, by a simple yet nontrivial transformation, it can be turned into a tractable form. As a result, an easy-to-implement conditional quantile estimation and inference procedure can be developed. Lastly, I will focus on the nonstationary vector autoregressive models with pure unit roots estimated by the LAD approach. A hybrid of the wild bootstrap and the randomly weighted bootstrap method will be proposed to approximate the asymptotic distribution of the LAD estimator, leading to three new bootstrapping panel unit root tests.
|Venue||Room 301, Run Run Shaw Building|
Registration is not required.
For further information, please visit:
Should you have any enquiries, please feel free to contact Ms. Irene Cheung by email at email@example.com or by phone at 3917 8312 or by fax at 2858 9041.