Dynamic panel data estimation, 2+ periods only for dependent variable

[Mary Jones]

Hi everyone,
I am trying to run a first difference regression which includes the lag of the dependent variable and other covariates. For all variables, I have data on the two relevant periods (t and t-1), however, for the dependent variable, I also have data on periods (t-2 and t-3). I would hence like to run a dynamic panel estimation xtabond or xtdpdsys, but this does not seem to work. For xtdpdsys I get estimates, but it says the number of instrument=2 even though I have 6 regressors.
For xtabond, I need to set artest(1) as an option (is this even valid??) and it says number of instruments=0.
If someone could point me into the right direction, I would really appreciate it.

I guess in a sense this is not a "dynamic panel" which is why this might not be working.
My alternative would be to use ivreg on the first differences and include all lags as instruments (and potentially the gmm suboption for efficiency?).
Would that be sensible?
Thank you!

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

You have too many different problems running around to get a good answer, but let me offer some comments. If you first difference your dv and have a lagged dv with two years of observations, you only have one usable observation at most. You can do a cross sectional sort of analysis, but are limited in being able to do most of the dynamic models.