Breusch-Godfrey after ARMA

[startz]

If you run a pure ARMA model and then check the Breusch-Godfrey, what exactly is EViews doing?

[startz]

bump?

[EViews Gareth]

Code: Select all

create u 100 series y=nrnd ' ar equation eq01.ls y c ar(1) eq01.makeresid resid01 eq01.makegrad grad01 grad02 series g1=grad01/resid01/(-2) series g2=grad02/resid01/(-2) series resid011 = resid01(-1) series resid012 = resid01(-2) resid011=@recode(resid011=na, 0, resid011) resid012=@recode(resid012=na, 0, resid012) equation serial_ar.ls resid01 g1 g2 resid011 resid012 freeze(artab) eq01.auto(2) ' arma equation eq02.ls y c ar(1) ma(1) eq02.makeresid resid02 eq02.makegrad grad03 grad04 grad05 series g3=grad03/resid02/(-2) series g4=grad04/resid02/(-2) series g5=grad05/resid02/(-2) series resid021 = resid02(-1) series resid022 = resid02(-2) resid021=@recode(resid021=na, 0, resid021) resid022=@recode(resid022=na, 0, resid022) equation serial_arma.ls resid02 g3 g4 g5 resid021 resid022 freeze(armatab) eq02.auto(2) show serial_ar show artab show serial_arma show armatab

[startz]

Hadn't seen that one before!
Thanks.

[kiber_master]

In such case the summary statistics of Breusch-Godfrey model will count intercept as another factor. I mean Adjusted R-squared and F-statistics, where k is the number of explanatories except intercept.