Dear all,

My panel is highly unbalanced, with total observations of 350 and lots of time gaps. These observations cover 20 years and 30 countries. For each country ranges, there are 1 to 20 data points.

Question 1: I tried Fisher type but failed. Stata shows "performing unit-root test on first panel using the syntax pperron variable, lags(2) returned error code 2001". Similar words appear when I change the options of dfuller, trend, drift, lags(2), demean. Is it impossible and unnecessary for me to conduct unit root tests considering my data condition?

Question 2: My panel is mainly constrained by the dependent variables because limited data are provided. I have more balanced and fewer gaps data for independent variables like inflation, export volume. For example, I have export volume data for 1990-2017 across more than 100 countries. Any suggestions under this circumstance?

Thanks a lot!

Rock