CAJ | 학술논문

现代金融风险管理中,利用条件异方差模型进行金融时间序列分析扮演着十分重要的角色,最典型的应用是基于观察到的离散时间价格或对数收益率序列做出管理决策.本文试图建立一种新的条件异方差模型去拟合资产收益率时间序列,该模型被本文提出的“正态分布的双变量连续二元正态混合(BNC-MN)分布”信息所驱动.该模型能充分捕捉资产收益率时间序列的“典型特征”如:非对称性,尖峰厚尾,波动群聚以及Black金融杠杆效应等；同时在此模型下,我们可以对上述特征做出较合理的经济学解释,并在一定程度上揭示用广义自回归条件异方差模型(GARCH)去模拟资产收益率序列的合理性.
현대금융풍험관리중,이용조건이방차모형진행금융시간서렬분석분연착십분중요적각색,최전형적응용시기우관찰도적리산시간개격혹대수수익솔서렬주출관리결책.본문시도건립일충신적조건이방차모형거의합자산수익솔시간서렬,해모형피본문제출적“정태분포적쌍변량련속이원정태혼합(BNC-MN)분포”신식소구동.해모형능충분포착자산수익솔시간서렬적“전형특정”여:비대칭성,첨봉후미,파동군취이급Black금융강간효응등；동시재차모형하,아문가이대상술특정주출교합리적경제학해석,병재일정정도상게시용엄의자회귀조건이방차모형(GARCH)거모의자산수익솔서렬적합이성.
Conditionally heteroscedastic models for time series play an important role in todays financial risk management,which typically tries to make financial decisions based on observed discrete time asset price and log-returns.This paper develops a new conditionally heteroscedastic model to fit asset returns,which is driven by bivariate normal continuous mixture of normal(termed BNC-MN) distributed innovations.The model can fully capture the stylized facts of asset returns time series,such as asymmetry,excess kurtosis and volatility clustering,even leverage effect.Meantime it can lend itself to reasonable economic interpretation of these stylized facts,and partly reveal the rationality of generalized autoregressive conditional heteroscedasticity (GARCH)-structure when employed to model asset returns time series.