GARCH模型的分位点回归的MCMC方法（英文）  

MCMC methods for quantile regression of GARCH models

作　　者：李东方[1] 唐亚勇[1] 

机构地区：[1]四川大学数学学院,成都610064

出　　处：《四川大学学报（自然科学版）》2016年第1期25-30,共6页Journal of Sichuan University(Natural Science Edition)

摘　　要：本文研究了GARCH模型和TGARCH模型的分位点回归的MCMC估计方法.通过将分位点回归估计问题转化为极大似然估计问题,并利用Metropolis-Hastings算法从模型参数的后验分布抽取随机数来完成GARCH类模型的分位点回归的Bayesian估计,本文得到了Value-at-Risk的动态估计.这一方法是对经典的Value-at-Risk计算方法的非常有效的推广.In the paper, we study the quantile regression estimation of GARCH models and TGARCH models, and propose the MCMC method （specifically, Metropolis-Hastings method） for the quantile regression estimators for the models. Based on the theoretic results, we carry out empirical investigation through the study of the return-on-assets of NASDAQ. And we obtain our method also provides a dynamic estimation of value-at-risk for the returns on assets, quantile regression of the models with the orooosed MCMC methods.

关 键 词：GARCH模型 TGARCH模型 分位点回归 MCMC Metropolis-Hastings算法 

分 类 号：O212.8[理学—概率论与数理统计]