ARMA model in panel data

[LEE CHAN]

Specifically, I have variable X representing the Invest/GDP ratio of 160 countries over a 22-year period. And I want to implement an ARMA model to consider the serial correlation of variable X. Is that possible? Can I run them all at once or do I have to split them into 160 regressions for each country?

[Andrew Musau]

What is your empirical model? How many other variables do you have?