Dear Zaki Farhana ,

Please share some additional information regarding your study - eg.

(1). the main objective of your study;

(2). specific variables of the model;

(3). economic theory underpinning your study; etc

Best

Dear Zaki Farhana ,

You can not go for a panel cointegration test with the I(2) dataset. It is required that all variables must be I(1) to apply the panel cointegration test.

I would like to suggest you transform the series into natural log form and then run stationarity test. You can also check for any structural break in your dataset.

Good luck.

Dear Zaki Farhana

I suggest that you first transform the series into growth rate or log form and then check the degree of cointegration.

Zaki Farhana, I am worried about the I(2) variable. you may do a log transformation and check the order of integration again. if ti is I(0) or I(1) then you can use it

In general, it is much better to test for cointegration even unit root says you have I(2) because it is possible that the I(2) results may result from other diseases such as measurement, absence of seasonality, outlier, etc therefore do not stop because another test may suggest the absence of I(2). By carefully studying your data you may come to a reasonable conclusion.