Page 1 of 1

t test statistics

Posted: Wed Mar 30, 2016 5:10 am
by samewing7
Hello,

I have an extremely basic question but its one i just want to confirm.

When I create a VAR I move to the estimation output page, which gives me the SE and t-statistics of the coefficients. What is the t-statistic testing? if it is significant, does that imply granger causality or that it is just a significant OLS estimator?

Thank for any help in advance.

Sam

Re: t test statistics

Posted: Wed Mar 30, 2016 9:01 am
by EViews Glenn
Standard t-statistic for the null that the coefficient is equal to zero.

Re: t test statistics

Posted: Thu Mar 31, 2016 3:33 pm
by dakila
Just ignore t statistics and coefficients . That does not make sense usually. Researcher never report VAR equation results but some test such autocorrelation and lag length. You should care about impulse response function and variance decomposition. I you wanna test granger causailty, use F statistic instead t statistic.