Hi everyone,

I am currently doing a dynamic panel regression analysis within the framework of my master thesis (n = 31 and T=17). I runned some unit root tests on all my variables and it seems that my index variable (0-5) does not pass the Fisher-ADF test with one lag but the IPS and Fisher-Phillips-Perron. Besides, my dummy variable (0-1) only passed all tests when I use the demean command. I wanted to first ask you if I have to run a unit root test on dummy variables at all? Since it seems that I only have stationary problems with dummy variables? Second, if I use the command trend and demean as well as 2 lags, my index becomes stationary (meaning that there exist a time trend and cross-sectional dependence), however I am not quite sure what do I have to do next with my model… Could someone may be indicate me the further steps to undertake in order to make my index variable stationary? Following that, my dummy variable (0-1) becomes stationary when I use the command demean. Once again, could someone help me figuring out the transformations to undertake? In that case, I wanted to run a cross-sectionally augmented IPS (CIPS) test by Pesaran (2007), but I am kind of lost regarding its use… Does anyone have an idea how to apply this test in STATA, especially regarding the use of/ difference between the options maxlags() and bglags()?

Thanks a lot in advance!!

Best

Ambre