EViews.com

hello,

I want to create a dummy variable for panel estimation. a dummy for the legal origin of the cross sections which are either french or english legal origin.

1- what are the steps to create dummy variable?
2- should I use one dummy variable or two for each category?
3- in case of one or two dummies; the coding of the dummy variable 1 and 0 or 1 for french and 2 for english for example?

Code: Select all

smpl @all series dum=0 smpl if legal="French" dum=1 smpl if legal="English" dum=2 smpl @all

Uh no. This is correct EViews code, but it's not a dummy variable. A dummy variable must be coded 0/1. You probably need two of them.

Yes, I just give an example to make a dummy with (0), 1, or 2 depending on the legal origin condition.

As an example, will produce a dummy with 1 if legal = English, 0 otherwise.

thank you,

i created two dummies one for french and one for english but i get an error message: near singular matrix??

Google "dummy variable trap."

Dear Startz

I have a question regarding the dummy variable in panel data.
I have a panel data which I want to estimate with PMG estimation. The friequency is weekly, however I included annual dummies, something like : my data is 1998 to 2014, so its 17 years weekly data and I created 17 dummy variables that takes value of one for each year and for all the cross sections. But the strange issue is that when I include the Whole annual dummy variables (17 ìannual dummies) it doesnt make any multicolinearity with constant, however when I do the same estimation with stata it drops the dummy of 1998, would you tell what is the reason for that result in Eviews?

Thank you

No idea. You might post your EViews workfile including the equation you are estimating.