EViews.com

Hi,

New here, great forum, very helpful. I have been trying to find ways of doing Sargan test or J-Test (I am under the impression they are the same, please correct me if I am wrong).

I am currently using EViews 6, I don't think my model would suit GMM, what is the other way of performing a test on the instrumental variables to see if they are suitable or if they are not too good, that is apart from J-test and Sargan test?

How would I perform these tests on EViews? Would I need to do more than pushing buttons (would be helpful if you could let me know which buttons to push!) or would I need to use equations (please provide equations again).

I would appreciate any help, I have searched google and everywhere for absolutely ages, but no good results have returned!

Many Thanks, very grateful!

Hi,

you just give the command:

scalar pval=@chisq(J-Statistic, instruemntrank-nummber of regressors in the modell). if your get a very small p-val it mean your model is bad. optimal value ist between 20- und 30%= 0.20 uns 0-30.

Thanks Hannibal! Very helpful!

I ran a variety of models, some come back saying that I need random effect, some saying I am fixed effect. For my fixed effects model, all of my instruments have p-value of 1.0000, whilst in random effects model, the Sargan p-value 0.0000

Confused!!

hi,
the p-val indicate, that you have misspecification in your model. maybe this due to high degree of autocorrelation!

Thanks Hannibal, very grateful, I will see what I can do! Cheers

Hi,

you just give the command:

scalar pval=@chisq(J-Statistic, instruemntrank-nummber of regressors in the modell). if your get a very small p-val it mean your model is bad. optimal value ist between 20- und 30%= 0.20 uns 0-30.

Hello
I did this command but the software doesnt run it and gave the error:statistic not defined.How i can check the problem and do sargan test in eviews.
thank u

You might want to show us the exact command you typed.