Hi, I have a small sample panel data consists of T=21 years and N=7 countries, with heterogenous slops and cross section dependence of my y and most of my x's . I did the unit root test my variables are mixed stationary I(0) and I(1). What is the best modeling method? Thanks for your replies.