EViews.com

Hi

I' m trying to understand the formulae behind the calculation of automatic bandwidth with kernel. When I work with Andrews method everything is ok but when the chosen method is Newey West (page 781 user's guide II) I'm facing some difficulties. I have a lot of work done but I'm afraid I'm a bit lost.
I'm sending you a simple model with only one parameter where you easily can follow my reasonning.
According to the guide I started calculating the various hypotesis of scalars autocovariances depending on the lags.All this calculations awhere based on the product sal*error ( independent variable *residual of regression) according to the model.
As in the example lag lengh = 1 it means that the autocovariances to use in forming the non parametric estimates f(q) are limited to this number, so it seems to me I should consider only 2 autocovariances (+1 and -1)in he calculation of f(q) with q=1 because of kernel function.
As to f(0) I considered the autocovariance for lag=0.

But the result Is not the one I should get, something is not correct or is missing, and I much appreciate if someone could give a help.

Thanks

João Pereira

Hello Gareth

Could you give me some tips about my issue?
I've been analysing some calculations of automatic bandwidth Newey West procedure but none of them match the result. I even don´t know if there are different methods we can use, or if the methods are simply not correct.The autocovariance formulae seem to be the same but perhaps there is something more to consider.Honestly speaking I' m a bit confusing with the subject.

I´d appreciate if you could give some support

Joao Pereira

Unfortunately, I think creating Excel files that compute econometric techniques it is a little outside of the scope of EViews support.

Hi Gareth

I´m sorry about that. I didn't know.

Thks anyway
J Pereira