Selecting data within the series - EV8

[H_Boss]

Hello,

I have series with multiple countries. Is it possible to filter 1 or several countries on estimation?


Thanx,

HB

[EViews Gareth]

Probably, but without more details it is impossible to give advice.

[H_Boss]

I wish to do the estimation with only 1 or some o'the countries in the series.
Is there a command like in SQL, "where country=Austria", f.e.?


HB

[EViews Gareth]

Set the sample to be

Code: Select all

If county = "Austria"

[H_Boss]

? on the estimation equation?

[EViews Gareth]

Yes

[H_Boss]

Error... "Near singular matrix."

Is this correct?
1980 2012 if country = "Austria"

[EViews Gareth]

That's correct.

[H_Boss]

Why the error then?

[startz]

Do you have a variable with constant value for Austria?

[H_Boss]

No, the values for Austria vary over the years 1980-2012. As for other countries.

[startz]

You might want to form the correlation matrix of all your right hand side variables, keeping the sample to Austria. Or you might want to post your workfile to see if anyone else sees anything.

[H_Boss]

Here it is...

[startz]

Please remember that while you're deeply involved, no one else understands the context. You need to tell us what equation you estimated, what options you used, etc.

[H_Boss]

Equation: logco2 c logres loghydro logfossil lognuclear loggdp logindustry urbpop
Method: Least squares, 1980 2012
Fixed effects, GLS: No weights.
Coef covariance method: I'm still trying to find that out (diff. post), using "Period weights" for now.