Page 1 of 1

hetroscedasticity in VAR model

Posted: Sat Jul 05, 2014 3:58 am
by adamalafensh
Hi

i am working on the return of three indices and their daily standard deviation generated by garch (1,1) with VAR model this including two group of dummy variables. so i have 6 variables overall
my returns are daily return with 530 0bservation. when i do run VAR Model, i do diagnostic test on residuals.
there is no serial correlation, however my model suffer from hetroskidasiticity.

this will bias my result in order to use the t statitics generated by the model. i will be more than happy to hear some advice of how i can deal with hetroscedasticity.

i obtaine 6 ADL equations, can i say that after obtaining the critical values from VAR, to run each ADL equation separately and to estimate it with robust least square ?

thank you
adam

Re: hetroscedasticity in VAR model

Posted: Sat Jul 05, 2014 10:58 am
by adamalafensh
?

Re: hetroscedasticity in VAR model

Posted: Sun Jul 06, 2014 1:41 pm
by EViews Glenn
Not sure that I fully understand what you are doing, but in general, you can estimate the VAR equation by equation using weighted least squares.