Page 1 of 1
J statistics and GMM
Posted: Sat Aug 16, 2014 8:10 am
by BGstudent
Hello,
When EViews reports results from GMM estimation the "J-stats" is the J statistics right, not the probability of "J statistics".
Radoslav
Re: J statistics and GMM
Posted: Sat Aug 16, 2014 8:59 am
by EViews Gareth
Correct
Re: J statistics and GMM
Posted: Mon Aug 18, 2014 5:56 am
by BGstudent
Hi Gareth,
Does EViews have a feature which shows the probability of J-stat?
Radoslav
Re: J statistics and GMM
Posted: Mon Aug 18, 2014 7:54 am
by EViews Gareth
By default the prob is shown:
Re: J statistics and GMM
Posted: Mon Aug 18, 2014 8:21 am
by BGstudent
So it doesnt show it when the J-statistics is very low?
Re: J statistics and GMM
Posted: Mon Aug 18, 2014 8:26 am
by startz
I believe that you have a just-identified system, so the J-statistic is zero except for rounding error. The p-value doesn't mean anything.
Re: J statistics and GMM
Posted: Mon Aug 18, 2014 8:41 am
by EViews Gareth
You have a panel equation. We don't show the p-value on a panel.
Re: J statistics and GMM
Posted: Mon Aug 18, 2014 8:54 am
by BGstudent
Thank you, Gareth.
Re: J statistics and GMM
Posted: Thu Aug 28, 2014 1:04 pm
by delcredere
to get the pvalue of the j statistics you use the below:
scalar pval=@chisq(J-Statistic, instruemntrank-nummber of regressors in the model)
my question is how do you execute that command in eviews 8