EViews.com

]Hello all

I am currently working with a VECM model.
I am exploring Granger/Wald Causality and Pairwise Causality, however, there is some confusion.

If you look at picture 1, the Granger/Wald Causality test, you can see 4 variables have a significant p-value, meaning those variables have short-run casualty with the dependant variable. Also, jointly they share short-run causality with the dependant variable.

If you look at picture 2, the Pairwise Causality test, only one variable has one-way causality.

Residual and CUSUM tests are ok.

Questions:

1) Why do the Granger/Wald Causality test and Pairwise Causality show different results?
2) What are the problems with having different results?
3) Are these results ok to report?

Thank you

If your sample size is large enough for johansen co-integration test, because it has asymptotic properties and are not right for small samples. And I am not an expert, so don’t precisely know what the required sample size for vecm is.

I would use the results from the vecm? Because pairwise granger causality only looks at pairs? While vecm is estimated in a system, thereby the coefficients are significant or not significant in contrast to granger causality.

Hello
I have annual series of 38 observations for my VECM.

So you wouldnt use the pairwise?

My confusion is why they show different results.
i guess they are different tests?

cheers