Hello,

I have an unbalanced panel data set consisting of N=25,102 and T between 1-35 years (on average T=9 years) and would like to perform an ARIMA analysis. I have run Fisher-type unit root tests as these are the only tests that allow for unbalanced panel data, as far as I understand. The four tests all reject the null hypothesis of "All panels contain unit roots" at the 1% level. My question is what does this tell me about stationarity? It could be the case that only one or only a tiny share of the panels are stationary. Isn't the result trivial? Is this result sufficient to assume stationarity in my panel?

Thank you in advance for your help,
Mia