Jarque-berra Test For Normality

JB检定如果拒绝Normality,

表示模型设置有问题 (specification error),

可能遗漏了某些Variable或用错了模型.

ex : MSI(2)H-AR(3) model
===========  Multivariate analysis of residuals - Jarque Berra ========

Jarque and Bera Test, Under Ho : 'Residuals are normal'
JB stat:  2.296
Pvalue :  0.317
The Normality hypothesis cannot be rejected
========================================================================

MSVAR_JarqBerra():Jarque and Bera Multivariate Test of normality for residual
                  analysis purpose,adapted courtesy from Michaux (2004)

Hamilton(1989) model 是MSM(2)-AR(4)

文献模型是MSM(3)-AR(2)

取模型残差做检验

这在e-views很简单

只要按下Stats Table,

就可得到底下类似结果.

    OILRET

Mean             0.575839

Median           1.064325

Maximum         13.42101

Minimum        -24.88856

Std. Dev.        4.728202

Skewness        -1.270786

Kurtosis         8.667291

Jarque-Bera    300.5852

Probability    0.000000

Observations   187

********

所以Jarque-Bera就跟mean,median,..一样,非常基础