marburg 发表于 2010-1-17 06:29 
相对于ARCH而言,GARCH还有一个parameter parsimony的优点.比如,如果单用ARCH模型,你可能要用ARCH(5)或者更高阶,但是用GARCH的话,GARCH(1,1)可能跟ARCH(5)的效果差不多~~
请问能帮我解释一下，为什么GARCH(1,1)可能跟ARCH(5)的效果差不多呢？
In fact if you work on your math a little you will find that GARCH(1,1) can be written as an infinite order of arch model(recursive substitution). It is much similar way of ARMA model in a time series.
In your(real) case you may observe that the effects is close to zero after oder of 5. It should be a test available to tell you which model is better. For example, LR test.
扫码加好友，拉您进群
全部版块
我的主页
收藏
