广义预测误差方差分解及其在证券市场冲击传递性研究中的应用

Generalized Forcast Error Variance Decompostion and it′s Application in Studying the Shocks’ Transmission in the Security Market

方差分解是向量自回归模型中研究各变量的冲击对所有内生变量预测误差贡献的方法。文章介绍了广义预测误差方差分解,同传统的正交预测误差方差分解相比,这种方法的特点是不受向量自回归模型中变量排序的影响。文章利用广义方差分解研究了沪市各个分类指数之间的关系,显示了系统冲击在各个行业指数之间传递的特点。

Variance decomposition describes the contribution of every variable’s shock in the VAR model to the endogeneous variables’ forcast error variance.The paper introduces the generalized forcast error variance decomposition.Compared to the traditional forcast error variance decompostition,this new method doesn’t depend on the order of the variables in the VAR model. Using the new method,the paper studies the interrelationships between the five ShangHai sector indexes.The results suggest that the shock to one index not only bring volatility to itself,but also to other indexs;the results also explain the direction & extent of the shock’s transmission,show the characteristics of the indexes’ volatility.