Granger test in cointegrated VAR

[Sraffa]

Hi

I'm now estimating VAR with cointegrated variables, variables of I(1). Some other studies use differenced variables, I(0), to the Granger causality test in VAR even if the variables are cointegrated. Is it OK? I fear that the long term relation of the variables are not to be constructed by differenced variables, so that Granger causality cannot be interpreted.

Sincerely

[alnassero]

Hi

I'm now estimating VAR with cointegrated variables, variables of I(1). Some other studies use differenced variables, I(0), to the Granger causality test in VAR even if the variables are cointegrated. Is it OK? I fear that the long term relation of the variables are not to be constructed by differenced variables, so that Granger causality cannot be interpreted.

Sincerely

Dear,

I think that we are using the same methodology. I am examining the relationship between economic growth and different financial development indicators using panel data.I have one dependent variable (economic growth) and six independent variables (financial development indicators). I have used the panel unit root test and I found that the series is nonstationary; Thus, I have difference the series and test for cointegration. My results indicate that there is a long run cointegration relationship between economic growth and financial development indicators.

Now if I want to test for Granger-causality test within an error correction model (ECM) using Eveiws 6.0. How i can get the error correction term, test for the sig of the term, and estimate the Granger-causality test within an error correction model.