基于常数和门限AR-TGARCH模型的CAViaR研究  

The CAViaR research based on constant and threshold AR-TGARCH model

作　　者：简志宏[1] 彭伟[1] JIAN Zhi-hong PENG Wei(School of Economics, Huazhong University of Science and Technology, Wuhan 430074, Chin)

机构地区：[1]华中科技大学经济学院,湖北武汉430074

出　　处：《管理工程学报》2017年第2期200-208,共9页Journal of Industrial Engineering and Engineering Management

基　　金：国家自然科学基金资助项目(71171090)

摘　　要：本文对AR-TGARCH模型进行了改进,提出门限I-AR-TGARCH模型、门限II-AR-TGARCH模型以及常数-AR-TGARCH模型,常数-门限I-AR-TGARCH模型和常数-门限II-AR-TGARCH模型,并且对中国四个股票指数2006年到2014年的数据进行实证分析,研究结果表明门限II-AR-TGARCH模型,常数-门限I-AR-TARCH模型和常数-门限II-AR-TARCH模型均比AR-TGARCH模型要优越,且常数-门限II-AR-TGARCH模型是最好的,常数-门限系列模型均优于其相对应的门限系列模型,四个指数均受到滞后风险的影响,且中小板指数和深圳成指所受到的滞后风险较小,上证指数同深圳成指具有传导双向关联性,上证指数和深圳成指会对中小板指数产生影响,上证指数对台湾加权指数具有传导性,台湾加权指数对上证指数却没有因果传导关联性。The outbreak of financial crisis has a huge impact on the global economy. The Asian financial crisis which broke out in 1998 had affected the economy of the entire Asian region, including Hong Kong and Taiwan of China. The outbreak of the U.S. subprime mortgage crisis had a huge impact on our trade economy. It had a long term impact on our economy and financial volatility. It is important to understand how to measure and manage related risks. This paper uses threshold function and constant term to establish a few models, including threshold I-AR-TGARCH model, threshold II-AR-TGARCH model, constant-AR-TGARCH model, constant-threshold I-AR-TGARCH model and constant-threshold II-AR-TGARCH model. These models are used to analyze the data collected from the 2006-2014 Taiwan weihted index, Shanhai comt3osite index. Shenzhen comoonent index and the small board index. Wecompare the advantages and disadvantages of each model through DQ test, RQ values and LR statistic. The Granger causality test is used to analyze the four indices. In the first part, the introduction and literature reviews are introduced. The quantile regression methods are used to show the development of CAViaR models. The proposed improvement and innovation of CAViaR models are introduced in this part. In the second part, specific content and improvement of CAViaR models are introduced. DQ test, RQ values and LR statistics are used to evaluate the models. In the third part, we use CAViaR models to study four indices. We also use granger causality test to analyze the causal relationships between the four indices. The fourth part summarizes the advantages and disadvantages of these models and the causality test of the four indices. In summary, the results show that the four indices are affected by lag risks, Shenzhen component index, and the small board index. Our proposed models, including threshold II-AR-TGARCH model, constant-threshold I-AR-TGARCH model, and constant - threshold II-AR-TGARCH model, are better than AR-TGARCH model. Constant-threshold

关 键 词：条件自回归分位数 门限函数 AR-TGARCH模型 

分 类 号：F830.91[经济管理—金融学]