Hello,
I'm trying to run a very simple DFM but I'm having some issues. Here is the task.
I have two non-stationary series (say GDP and GDI), specified in logs (log_GDP and log_GDI). Each series is assumed to have a (common) non-stationary trend and an error.
I would like to run a DFM to estimate a common non stationary trend. The trend should be specified as an I(2) (smooth trend).

Therefore, the model should be of the form:

log_GDP = y + epsilon_GDP
log_GDI = y + epsilon_GDI

where y = x + epsilon_y
and x is a random walk.

How do you estimate such a model in STATA?
(I tried but it seems I can only specify y as an AR(2), and the model has typically convergence issues).
Thanks,
R.