So I have some panel data, investment in different regions over the course of a 7 years. It's non stationary, and I'm hoping to use it in a Pooled OLS, as my dependant and other independent variables are all stationary.
Here is a Levin Lin Chu result : https://i.imgur.com/eAmB9v6.png
Here is a scatter of the data : https://i.imgur.com/5r4lS87.png
It's currently in real terms, not nominal, but there is still a deterministic upward trend. Is it valid to just log the data and use that variable? Or would it be more statistically sound to detrend it? Does anyone have any book or paper recommendations to guide me, because I'm pretty confused.
EDIT :
I've tried logging the variable and it does not remove all of the non-stationary. Although lagging the variable by 3 years does, unfortunately lagging the variable by this much shows weakened results in Aike Information Criterion tests and in regression results.
Here is a Levin Lin Chu result : https://i.imgur.com/eAmB9v6.png
Here is a scatter of the data : https://i.imgur.com/5r4lS87.png
It's currently in real terms, not nominal, but there is still a deterministic upward trend. Is it valid to just log the data and use that variable? Or would it be more statistically sound to detrend it? Does anyone have any book or paper recommendations to guide me, because I'm pretty confused.
EDIT :
I've tried logging the variable and it does not remove all of the non-stationary. Although lagging the variable by 3 years does, unfortunately lagging the variable by this much shows weakened results in Aike Information Criterion tests and in regression results.
