Unit root test on panel data

[Joyce Parreira]

Hi!

I have a panel data with 155 countries and 43 years.

I tried the command

xtunitroot ht lngdppppcte2011

but ir returns that "Harris-Tzavalis test requires strongly balanced data"

I also tried this

xtunitroot fisher lngdppppcte2011, dfuller lags(1)

and worked.

Is there any big difference between them?

Moreover, I'm regressing all my independent variables lagged by 1 period. But the unit root teste should be on the contemporary variable or can be on the lagged variable?

Thank you so much!