Combining time series - Granger tests

[Tommaso Felici Netherlands]

Hi!

I would like to combine more Granger tests to understand whether there is causality between two variables in different units.

In other words:
let's imagine that I have 3 countries, and I am interested in the causality from X to Y.
I would like to run three different granger tests for each country, collect the three p-values and then "combine" them to obtain an "average overview" of the casualty from X to Y.

Is it possible? And how?

Thank you very much!

[Scott Merryman]

You can use Fisher's method to combine multiple p-values into a single test: https://en.wikipedia.org/wiki/Fisher%27s_method