Fixing a variable to its initial value

For requesting general information about EViews, sharing your own tips and tricks, and information on EViews training or guides.

Moderators: EViews Gareth, EViews Moderator

rmitri
Posts: 2
Joined: Fri Feb 17, 2017 6:48 am

Fixing a variable to its initial value

Hi there,

I'm doing my master thesis and studying the impact of debt on economic growth.

I'm working on a panel data of 12 european countries during the years 1991 - 2015.

I would like to set the GDP variable to its initial value (value in 1991) in order to test for conditionnal convergence across economies (higher (lower) initial value of GDP tends to lower (higher) economic growth). Do you know how to procede ?

Here is the equation I want to estimate :

y i,t = c + β0 lnGDP i,0 + β1 debt i,t + β2debt2 i,t + β3 gfcf i,t + β4 popgrowth i,t + ui + vt + Ɛi,t

But I can't find a way to fix the lnGDP variable to its initial value.

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 11730
Joined: Tue Sep 16, 2008 5:38 pm

Re: Fixing a variable to its initial value

Something like:

Code: Select all

`series fixedgdp = @meansby(gdp, @crossid, "1991 1991")`

rmitri
Posts: 2
Joined: Fri Feb 17, 2017 6:48 am

Re: Fixing a variable to its initial value

Thank you a lot !

It worked but I couldn't run the regression with fixed cross-section (near singular matrix error), do you know why ?

startz
Non-normality and collinearity are NOT problems!
Posts: 3299
Joined: Wed Sep 17, 2008 2:25 pm

Re: Fixing a variable to its initial value

Perfect multicollinearity. Each cross section already has a constant--that first observation.