基于TDAR模型的VaR估计方法及应用  

The Estimating Method of VaR Based on the Threshold Double AR Model and its Application

作　　者：蒋勇[1,2] 吴武清[3] 王力伟[4] 叶五一[1] 陈敏[5] 

机构地区：[1]中国科学技术大学统计与金融系,安徽合肥230026 [2]中国人民银行征信中心,北京100031 [3]中国人民大学商学院,北京100872 [4]中国银行风险管理部,北京100872 [5]中国科学院数学与系统科学研究院,北京100190

出　　处：《中国管理科学》2012年第5期1-6,共6页Chinese Journal of Management Science

基　　金：国家自然科学基金资助项目(70221001;70331001;10628104;71003100);中国人民大学科学研究基金项目(11XNK027;10XNF020);安徽省自然科学基金(090416245);高等学校博士学科点专项科研基金(20103402120010)

摘　　要：文献中,在险值估计方法一般基于线性假设,但是该假设在实际中很难满足,需要为此提出非线性的在险值估计方法。与以往传统模型一般假定变化发生在"时间"点上不同,门限双自回归(TDAR)因状态空间的不同而建立不同模型来对非对称性、结构变点等非线性现象进行刻画,并同时允许均值和波动率过程的结构变化。本文首次基于TDAR建立TDAR-VaR方法,并对上证指数和香港恒生指数进行了实证研究和对杠杆效应进行了分析。实证分析发现TDAR-VaR较好地预测了市场风险。In most literature, the measurement methods of VaR are based on the linear statement. Howev- er, in practice this assumption cannot be satisfied quite well. So it is needed to put forward the nonlinear estimation method of VaR. In contrast to the traditional model that allows model changes to occur in the ＂time＂ space, the threshold double AR model （TDAR） uses threshold space to model the nonlinear phe- nomena such as asymmetry and the structure change, and also allows the structure change of mean and vol- atility. In this paper, the method of TDAR-VaR is presented for the first time, and the empirical research and the leverage effect analysis on SZZS and HIS index are also covered. The empirical analysis shows that this method can predict the market risk very well.

关 键 词：门限双自回归模型 在险值 非线性 杠杆效应 

分 类 号：F830.9[经济管理—金融学]