how to remove heteroskedasticity in ARDL model
Posted: Sun Jul 14, 2013 5:01 am
Hi, i wonderg if anyone can help me... im regressing an ARDL model. My equation looks like
lib = a + lib(-1) + base(-1) + lib(-2) + base(-2) + ....
i estimated the optimal lag length looking at the SC & AIC criterion and both suggested that i should use ARDL3
My prob is that now as im running the residual diagnostic test .. in the heteroskedasticity test i have to reject null hypothesis ( i.e. there is homoskedasticity in my model). ive tried reading solutions on the web, some suggest to take logs which i cannot do because in my case my variables are interest rates (%) so i cant take logs as that would render them meaningless.
If anyone can offer a solution i would be really grateful.
Hira
lib = a + lib(-1) + base(-1) + lib(-2) + base(-2) + ....
i estimated the optimal lag length looking at the SC & AIC criterion and both suggested that i should use ARDL3
My prob is that now as im running the residual diagnostic test .. in the heteroskedasticity test i have to reject null hypothesis ( i.e. there is homoskedasticity in my model). ive tried reading solutions on the web, some suggest to take logs which i cannot do because in my case my variables are interest rates (%) so i cant take logs as that would render them meaningless.
If anyone can offer a solution i would be really grateful.
Hira