Hi!

I am using data from the British Household Panel Survey which is an unbalanced panel. I am primarily using categorical variables like whether the individual is self employed or not, part vs full time etc. I am using both fixed effects regressions exploiting the panel dimension but also i have taken yearly averages of some variables and ran time series regressions. I would like to test for the possibility that some of my variables have stochastic trends. However, when i do panel unit root tests (for instance Levin-Lin-Chu), i get a message saying 'repeated time values within panel'. When i do a normal ADF test on the yearly averaged variables used for the time series regression, i get that 'sample may not include multiple panels'.

Does anyone have any suggestions on how i could resolve this problem? Solving one of the two (meaning if i manage to test for stochastic trends either the yearly averaged variables or the non-averaged ones will suffice) will be enough. Thank you in advance

Christos