Depends on the software you are using - R, E-views or Stata. E-views has specific guidelines which are very easy to follow with examples.

You can try these steps. Step 1: Testing for the Order of Integration. Panel unit root used (e.g. Maddala and Wu 1999),

Step 2: Testing for possible Cointegration. Pedroni test for cointegration used among variables.

Step 3: Estimating long run parameters (If Cointegration is found in Step 2). Long Parameters using DOLS method.

Step 4: Examining the causal direction a panel vector error correction model Pesaran et al. (1999) will be estimated. The Engle-Granger (1987) two-step procedure will be used.

Hope these steps help.

is it help?

You can watch video tutorials, which are available in youtube, specially in Hossain academicy channel.

Dear Elfan,

thank you very much. I was able to rebuild the examples. Based on the examples, I am able to do the "Granger test" with my own data.

Dear Bokhitar,

I had a look to Hossain academicy channel (e.g. Granger Causality Test. Model One.Part 1 of 2. EVIEWS). The tutoralias are very helpful.

Thank you !!

You are welcome. And all the best.