Recursive residuals and dummy variables

[MOG]

Hi,

It seems it is not possible to get recursive residuals for the full sample when a dummy variable is inlcuded in the equation. Why is that?

Regards
Martin

[EViews Gareth]

I'm not sure I understand the question. This seems to work just fine:

Code: Select all

create u 100 series y=nrnd series dum = nrnd>0 series x=nrnd equation eq1.ls y c x dum eq1.rls(r,s)

[MOG]

Yes, you're right. But for this case it does not work:

Code: Select all

create A 1980 2010 series y=Nrnd series x=nrnd Series dum = 0 Smpl 2000 2010 Series dum = 1 Smpl @All equation eq1.ls y c x dum eq1.rls(r,s)

[EViews Gareth]

Any sample ending before 2000 will be singular due to perfect collinearity.

[MOG]

Yes, of course, since in the smaller samples the dummy-variable values are constant, and the matrix of varible-data has reduced rank until the sample size is big enough to allow the dummy-variable to switch values. Thanks!