dummies

[stefan]

LS,

I hope someone can help me I've been trying to get run a OLS regression, with about 18 variables, in eviews 6.0.
Of the variables about 11 have to be dummy variables, I've been trying to get eviews to generate these for days and it's driving me crazy.

The dummies have to be for things like:
geographical (USA, EU, Other, World)
Publication (AA, A, B, C, N)
time periode analysed (in decades)
( thus the dummies need to be such that multiple can be used in one regression equation)

I use Eviews 6 standard edition, which says it has a 'Automatic Categorical Dummy Variables' function that should create these dummies by @expand()
C:\Program Files\EViews6\Help Files\wwhelp\wwhimpl\js\html\wwhelp.htm
However when I try to create these dummies :
with Genr (in the little window with my variables) and than (by generate series by equation) enter @expand(geographical) it won't work
it says: @EXPAND is not a genr or series expresion function

or when i try in an equation eg: effect c years_post observations @expand(geographical) than it gives syntax error

The first is the way i would like to create the dummies, as they than can be used indiviualy.

can someone please help me out ?
I can't find anything on the net.

I'm quite new to eviews to please go step by step.

Thanks in advance for saving the day (or week).

Stefan

[EViews Gareth]

Type:

Group G @expand(geographical)



If that doesn't work, post your workfile.

[stefan]

Dear QMS Gareth,

It seems to have worked! Thank you very much for the quick reply !

Should it now possible to use the new group G straint in to a regression? say y = b1 x1 + b2 x2 + b3 G ?
Because it than gives a "Near Singular Matrix" warning.

(I read the C:\Program Files\EViews6\Help Files\wwhelp\wwhimpl\js\html\wwhelp.htm page but that didn't clear things up)

ps can you tell me whan is the difference / when do you use the 'genr' button in the workfile window and when do you type in the white bar above?

Thanks again

[startz]

Dear QMS Gareth,

It seems to have worked! Thank you very much for the quick reply !

Should it now possible to use the new group G straint in to a regression? say y = b1 x1 + b2 x2 + b3 G ?
Because it than gives a "Near Singular Matrix" warning.

99 % chance you've included a full set of dummies plus a constant, which can't be done.

[EViews Gareth]

Dear QMS Gareth,

It seems to have worked! Thank you very much for the quick reply !

Should it now possible to use the new group G straint in to a regression? say y = b1 x1 + b2 x2 + b3 G ?

Well not quite, since G is a group of variables, you can't have it multiplied by a single coefficient (b3 in your notation).

Because it than gives a "Near Singular Matrix" warning.

As startz pointed out, almost certainly because you've fallen into the dummy variable trap.

ps can you tell me whan is the difference / when do you use the 'genr' button in the workfile window and when do you type in the white bar above?

Thanks again

There is none.

[stefan]

Thanks for the reply guys!

was away, so a little late response.

Gr.