基于TGARCHSK模型的外汇市场极端风险测度研究  

Extreme Risk Measurement of Foreign Exchange Rate Market based on Time Varying High Order Moments

作　　者：吕永健[1] 王鹏[1] 胡颖毅[1] 

机构地区：[1]西南财经大学中国金融研究中心,成都611130

出　　处：《系统管理学报》2017年第2期234-244,共11页Journal of Systems & Management

基　　金：国家社会科学基金重大项目(13&ZD030);国家自然科学基金面上项目(71473200);国家自然科学基金资助项目(71101119);西南财经大学和四川省教育厅创新团队建设项目(JBK130401);中央高校基本科研业务费专项资金资助项目(JBK1507085)

摘　　要：针对金融资产收益的时变高阶矩波动特征在风险管理中的作用和意义,提出了一个新的时变高阶矩波动模型——门限广义自回归条件异方差-条件偏度-条件峰度(Threshold Generalized Autoregressive Conditional Heteroskedastic,Skewness and Kurtosis,TGARCHSK)模型,并运用后验分析方法,考察了基于TGARCHSK模型与基于常数高阶矩波动模型,以及其他时变高阶矩波动模型的国际外汇市场ES风险测度精度差异,得出:(1)在将条件偏度和条件峰度纳入建模框架之后,条件方差的波动幅度将会变小;(2)在ES风险测度中,时变高阶矩波动模型的精确性要显著优于常数高阶矩波动模型;(3)各模型的风险测度精确性在不同头寸下会表现出显著的差异;(4)通过综合考虑不同模型在后验分析中表现出的精确性差异,认为对于精确测度外汇市场的ES值,TGARCHSK模型可以作为相对合理的模型选择。Recently, more and more research begins to focus on the role of Skewness and Kurtosis in riskmanagement. This paper extends the research of Leon （2005）, and put forward a new time-varying higher momentvolatility model, namely, Threshold Generalized Autoregressive Conditional Heteroskedastic, Skewness andKurtosis model. Based on the constant higher moment volatility model and time-varying higher moment volatilitymodel, we study the ES measurement of the international foreign exchange market. The study shows （1） similarto Harvey （1999）, when incorporating the conditional skewness and Kurtosis into the modeling framework, thefluctuation range of conditional variance become smaller; （2） the accuracy of time-varying higher order momentsvolatility model is significantly better than the constant higher moments volatility model on the measurement ofES; （3） the accuracy of risk measurement of each model is obviously different in different positions; （4） combiningall the facts, TGARCHSK model can be used as a relatively rational choice for estimating the ES of theinternational foreign exchange market.

关 键 词：金融风险测度 预期损失 国际外汇市场 后验分析 

分 类 号：F830.92[经济管理—金融学]