Hi,
I was trying to replicate the results obtained by xtabond with the new
xtabond2, but did not succeed
For example, using the Arellano-Bond (1991) data, I was thinking that
the following commands yield that same results:
webuse abdate, clear
xtabond n l(0/1).w l(0/2)(k ys) yr1980-yr1984, lags(2) noconstant
xtabond2 n l(1/2).n l(0/1).w l(0/2)(k ys) yr1980-yr1984, gmm (l.n w k
ys) iv(yr1980-yr1984) noleveleq
Since the following commands yield the same coefficients, I must be
somehow wrong on the instruments, I guess:
xtabond n yr1980-yr1984, lags(2) noconstant
xtabond2 n l(1/2).n yr1980-yr1984, gmm (l.n) iv(yr1980-yr1984) noleveleq
However, the standard deviations slightly differ. Does anyone know the
reason? In the twostep case, they are equal.
And, as a last question, is there a way to include a constant into the
first-difference equation of xtabond2?
