求助推倒Bartlett kernel 的optimal bandwidth

cbqywl 发表于 2013-9-11 13:30
The paper I mentioned is "Heteroskedasticity and Autocorrelation Consistent Covariance Matrix Estima ...

I have read it, in that paper, he used QS kernel as optimal one not Bartlett. But the method is the same. I I only let residual subjected to AR(1) with autoregressive parameter equal to theta. but I dont know how to calculate this complicated formula. I have plug-in all the parameter but you know, my maths like rubbish. ....  Where are you right now? in US or UK?

你好，我现在在澳洲，办公室电脑没法打中文。。。对于AR MA模型 那一项是可以算的，对于AR模型 那一项由innovation的方差和自回归系数决定。如果我没记错的话 Andrews的文章有给出AR模型那一项的的具体值。当然你也可以自己计算，因为AR模型的不痛滞后阶的协方差是可以算的，将那些协方差代入那个公式，求无限个协方差的和。