What to do with non-stationary and not cointegrated panel data?


New Member

I've been trying to conduct a panel data analysis. I have 3 variables, all of them stationary at I(1). Then I employed both Westerlund and Pedroni panel cointegration tests. Both tests gave the same result, my data are not cointegrated. And now i am lost, don't know what's next? How can i proceed my analysis? Should i go on with first differences and just use random or fixed effects models?
I'll appreciate any help...