How best can I use the Granger causality test in time series data and understand it better because I have never used it. I want to analyze long run relationship and bi-directional relationship between two variables.

  • $\begingroup$ @Andy edited in the granger-causality tag, which shows that there are lots of questions on the topic on this site. Have you looked at any of them? Are you sure you have a new question? Have you studied basic texts and internet sources? $\endgroup$
    – Nick Cox
    Jun 7, 2014 at 10:10
  • $\begingroup$ from the few sources I have looked at, the model seems to be written different and the are barely describing the letters on the model $\endgroup$
    – Daniel
    Jun 7, 2014 at 10:20
  • $\begingroup$ I don't think I understand that comment. However, you seem to be asking for a personal tutorial starting from scratch on what in its field is an utterly standard topic. You may be lucky, but a common attitude from many people here will be "Please do read a textbook at your level". $\endgroup$
    – Nick Cox
    Jun 7, 2014 at 10:24
  • $\begingroup$ will try to get to books thou I just wanted a little understanding especially on interpreting the symbols on the model $\endgroup$
    – Daniel
    Jun 7, 2014 at 10:34
  • $\begingroup$ Daniel, why don't you try writing out the model (using the notation from the texts) and ask specific questions about the symbols you are puzzled by? $\endgroup$
    – Alexis
    Jun 7, 2014 at 16:07

1 Answer 1


Here is an off the cuff answer.

Granger causality means that the shock or error from one variable has a cross correlation or impacts another.

H0 or the null hypothesis is NO CAUSALITY. It means your result falls within a range or confidence interval close to zero significance.

If the p>.05, the probability you 'struck out' and the one variable does not impact the other is high enough you 'cannot reject the null of no causality.' Stats-speak: you must accept that you failed to show causality on the variable.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.