Page 1 of 1

least square dummy variable with restrictions

Posted: Mon Feb 01, 2010 8:40 pm
by simplesemplice
Hi all,

I understand that Eviews cannot set restricts on coefficients, but I would like to construct a least square dummy variable model, LSDV that includes the intercept and all dummies and then imposing a restriction on the model.

My equation is as follows:
atr = constant + lgdpf + irf + vrf + dum_na + dum_sa + dum_us
I understand I need to restrict coefficients of my 3 dummy variables to be equal to zero.

Can someone help, as to how I can run this regression, without being in the dummy variable trap? much thanks!

I've attached my file.

Re: least square dummy variable with restrictions

Posted: Wed Feb 03, 2010 10:57 am
by EViews Gareth
Just re-paramaterise your model so that the coefficient on dum_us become (0-Cdum_na-Cdum_sa) where Cdum_na is the coefficient on dum_na and Cdum_sa is the coefficient on dum_sa. Thus something like:

Code: Select all

ls atr=c(1)*lgdpf+c(2)*irf+c(3)*vrf+c(4)*dum_na+c(5)*dum_sa + (0-c(4)-c(5))*dum_us